Chapter 16: Toolmark and Impression Evidence: Footprints, Tire Tracks, and Bite Marks (the Debunked One)

"Bitemark comparison is an area of forensic odontology … the scientific basis for which has not been established." — paraphrasing the consensus that emerged from the 2009 National Academy of Sciences review and the 2016 PCAST report; the discipline that opens and closes this chapter is the one those reports treated most harshly.

Overview

A door is forced. A car waits in the mud at the end of a gravel track and then drives away. A burglar climbs through a window and crosses a damp flower bed. Each of these acts leaves the same kind of evidence: a harder object presses into, scrapes across, or cuts through a softer one, and the softer one keeps a record of the shape that touched it. That record is an impression or a toolmark, and reading it is one of the oldest crafts in forensic science. At its best, it is solid, useful, and honest about its limits — a tire that narrows the search to one tread design, a pry bar whose width and shape are consistent with the gouges on a doorframe. At its worst, it is the single most thoroughly debunked discipline the field has ever taken into a courtroom: bite-mark "matching," which has helped convict, and in some cases nearly execute, demonstrably innocent people.

This chapter is the validity spectrum made visible in one room. We start with the logic all impression evidence shares, then work through footwear, tire, and toolmark comparison — disciplines that range from genuinely valuable (footwear class evidence) to seriously contested (the claim that a single tool left a single mark to the exclusion of all others). Then we turn to bite marks, and we do not pull the punch: we will trace exactly how a method with no validated foundation became confident courtroom testimony, and we will meet two men — Ray Krone and Roy Brown — whom bite-mark "experts" helped put away for crimes DNA later proved they did not commit.

The lesson is not "impression evidence is junk." Most of it is not. The lesson is that the same family of evidence contains both careful, defensible work and pseudoscience wearing a lab coat, and that the difference is exactly what the 2009 NAS and 2016 PCAST reports taught us to ask about: is there a validated method, with a known error rate, for the specific claim being made? For a tire's class characteristics, broadly yes. For "these teeth and no others made this wound," emphatically no.

In this chapter, you will learn to:

• Explain the shared logic of impression and toolmark evidence, and apply the class-vs-individual distinction (Chapter 1) to each.
• Describe how footwear and tire impressions are documented, lifted, and cast, and state honestly what they establish.
• Evaluate toolmark comparison without overstating it, and place firearms/toolmark identification on the validity spectrum.
• Reconstruct the rise and fall of bite-mark analysis and name the unvalidated assumptions at its core.
• Connect the Krone and Brown exonerations to the book's four themes.
• Say precisely what an examiner can honestly testify to about each kind of mark — and what crosses the line.

Learning Paths

🔎 Investigator/CSI: This is a documentation-and-collection chapter for you. Sections 16.2 and 16.3 are your field manual — photograph before you cast, cast before the weather changes, and remember that a footwear impression you walk past is gone forever. The single most common avoidable error in this whole area is failing to recover the impression properly at the scene. 🧪 Lab analyst: Weight §16.4 and §16.5. The toolmark examiner's hardest job is telling a true individual mark from a subclass artifact (Chapter 1), and the bite-mark story is a cautionary tale about what happens when a comparison discipline never measures its own error rate. Your verbs matter here more than almost anywhere. ⚖️ Law/courtroom: Sections 16.4, 16.5, and 16.6 are your chapter. Toolmark "identification" is admitted in most courts but rests on contested ground (PCAST 2016); bite-mark testimony is now excluded or sharply limited in a growing number of jurisdictions. Know the difference and know the exonerations. 👥 General reader/juror: Read §16.1, §16.5, and §16.6 closely. A confident expert holding two photographs side by side is the most persuasive — and, for bite marks, the most dangerous — thing you may ever see in a courtroom. This chapter is your inoculation.

16.1 The logic of impression evidence

Begin with the question the evidence is being asked to answer. A house has been burgled; the back door is splintered around the lock. The investigator's question is not yet "who did this?" It is narrower and more answerable: was this door forced, and if so, by what kind of thing? That is a question about shape. Whatever pried the door left, pressed into the wood, a negative record of itself — the width of a blade, the curve of a hook, the flat of a heel. Impression evidence is physical evidence in which the shape of one object is transferred to, and recorded on, the surface of another. A footwear impression in mud, a tire track in snow, the tool-gouge in a doorframe, a bite in skin or in a half-eaten sandwich left at a scene — all are impressions, all governed by the same simple physics.

The physics is this: when a harder surface contacts a softer one with enough force, the softer surface deforms to record the contours of the harder one. The recording can be three-dimensional (a shoe pressed into soil, leaving depth as well as outline) or two-dimensional (a dusty shoe stepping onto a clean tile, leaving a flat transfer of the tread). It can be a positive impression — material deposited, like a bloody shoeprint on a floor — or a negative one, like the void a tire carves into mud. And critically, it is governed by the same comparison logic as every other discipline in this book, the logic we built in Chapter 1: every impression carries class characteristics, shared by a whole group of possible sources, and may carry individual (or "acquired") characteristics, the accidental damage that, if there is enough of it, can narrow a source toward one.

This is also Locard's exchange principle (Chapter 3) in one of its most legible forms. When the burglar's pry bar bit into the doorframe, the door took an impression of the bar — and, just as surely, the bar took splinters and paint from the door. Impression evidence is the door's half of that exchange, frozen in place; the trace transferred to the tool is the other half (paint and wood comparison, Chapter 24). Reading both halves is how a scene gets reconstructed.

🔬 At the Bench The whole of impression comparison is a disciplined version of something you already do without thinking. You recognize a friend's handwriting, a particular car's rattle, the worn spot on your own front step. The examiner formalizes that recognition: document the questioned impression from the scene, obtain a known impression from the candidate source (make a test print with the suspect shoe, a test mark with the suspect tool), and compare them feature by feature for agreement and, just as importantly, for unexplained difference. The single most important rule is symmetric with DNA (Chapter 7): a genuine, unexplained difference can exclude, while agreement only associates — and how strongly it associates depends entirely on how far along the class-to-individual arrow the agreeing features sit.

Two ideas from Chapter 1 do most of the work in this chapter, so we restate them in the language of marks. A class characteristic of an impression is a feature determined by the design or manufacture of the source: the tread pattern molded into every shoe of a model, the rib spacing engineered into every tire of a line, the width of a chisel as it left the factory. Class characteristics can be shared by thousands or millions of objects. An acquired or individual characteristic is a feature a particular object picked up through use and damage: a cut across one shoe's tread, a nick on one chisel's edge, a plug of gravel wedged in one tire. These are, in principle, random and effectively unique to the individual object — if there are enough of them and if they are reliably reproduced in the impression.

The honest examiner lives on a strict hierarchy of conclusions, and you should learn it now because it recurs for every impression type in this chapter:

• Exclusion. The known source could not have made the questioned impression (wrong size, wrong tread design, an unexplained difference). Clean and often the most defensible call.
• Class association ("could have made / is consistent with"). The questioned impression and the known source share class characteristics; the source — and every other object of its class — cannot be excluded.
• Association with individual characteristics ("identification" in some lab vocabularies). The impression shares not only class features but enough corresponding acquired damage that the examiner concludes the source made it. This is the contested top of the hierarchy, and where overstatement creeps in.
• Inconclusive / unsuitable. The impression is too faint, partial, or smeared to support any of the above.

Hold that hierarchy in mind. The entire validity argument of this chapter is about which methods can honestly reach which rung — and bite marks, we will see, routinely claimed the top rung from a foundation that barely supports the second.

🔍 Check Your Understanding 1. A shoe impression in mud shows both a tread design and a deep, distinctive cut across the sole. Which is a class characteristic and which is acquired? Which one, if matched to a suspect's shoe, narrows toward a single source? 2. Why can a single unexplained difference between a questioned and a known impression be more powerful than a dozen points of agreement?

16.2 Footwear impressions: class and acquired characteristics

Footwear impressions are, after fingerprints, perhaps the most common pattern evidence at scenes — and one of the most neglected. Almost everyone wears shoes; almost everyone who commits a crime in person walks to and from it. Yet footwear evidence is routinely overlooked, walked through, or lost to weather, because unlike a fingerprint it is large, fragile, and easy to destroy. The first lesson is therefore not about comparison at all. It is about recovery. A footwear impression you fail to document at the scene is not weak evidence; it is no evidence.

Documenting and recovering the impression

The order of operations is fixed, and it protects against destroying the thing you are trying to capture. First, photograph — overall shots to fix the impression's location, then a series of examination-quality photographs taken with the camera lens parallel to the impression, a scale (a ruler, ideally an L-shaped scale showing two axes) in the same plane as the print, and oblique lighting raked across the impression from a low angle to throw its three-dimensional detail into relief. The oblique-light photograph is the single most valuable record of a three-dimensional impression; tread detail invisible under flat light leaps out under a flashlight held low and to the side.

Only after photography does the investigator recover the impression physically, by one of two routes depending on its nature:

• A two-dimensional impression (a dust or residue print on a hard surface) is lifted — with an electrostatic dust-print lifter that draws fine particles up onto a charged film, or with a gelatin or adhesive lifter pressed over the print.
• A three-dimensional impression (a print in soil, sand, or snow) is cast. A cast is a three-dimensional reproduction of an impression, made by pouring a casting medium — typically dental stone, a hard gypsum product — into the impression and letting it cure into a positive replica of the shoe's underside. For impressions in snow, a colored wax (an aerosol such as the commercial product sometimes called snow-print wax) is sprayed in first to define and insulate the detail before the dental stone is poured.

FIGURE 16.1 — Recovering a three-dimensional footwear impression (schematic, not to scale)

   PHOTOGRAPH FIRST                       THEN CAST
   ----------------                       ---------
        camera ↓ (lens parallel)
                                          dental stone poured in,
     ▭ scale ──────                       cured ~30 min, lifted out
        ___________                            ___________
   ----/  tread    \----  ← oblique           /  POSITIVE  \   ← a 3-D replica of
       \  detail   /        light from        \   replica  /     the shoe's sole
        \_________/         a low angle         \_________/      (the "cast")
        (negative void in soil)

   Legend:  the soil holds a NEGATIVE of the sole; the cast is a POSITIVE you can
            compare directly to a suspect shoe. Measurements illustrative.

The prose walkthrough of Figure 16.1: the impression in the soil is a negative — where the sole stuck out, the soil is pushed in. Dental stone poured into that void cures into a positive — a hard replica of the bottom of the shoe, which the examiner can handle, measure, and lay beside a test impression made with the suspect's actual shoe. Casting is slow and unforgiving (a cast poured too thin, or lifted before it cures, can shatter and take the only record of the impression with it), which is exactly why photography comes first and is never skipped. The photograph is the backup; the cast is the prize.

🔎 Investigator/CSI — field note The classic footwear error is the failure to look down. Responders and investigators tramping across a scene leave their own impressions and obliterate the offender's. Discipline at the scene — a single controlled path of entry, eyes on the ground, every impression flagged and photographed before anyone collects anything — is worth more than any laboratory technique that follows. You cannot analyze what you stepped on. (This is the same scene-integrity discipline you met in Chapter 2; impression evidence is where carelessness shows up fastest.)

What footwear comparison establishes — and what it doesn't

Now the comparison. A footwear impression is the recorded outline and tread detail left by the underside of a shoe, and it carries the full class-to-individual range:

• Class characteristics: the tread design (the molded pattern), the size, the general wear pattern (how the wearer's gait wears the sole — outer edge, heel, ball). Tread design and size together can be quite discriminating as class evidence: a particular outsole pattern is made by one manufacturer in a finite run of sizes, and databases of footwear outsole patterns let an examiner identify the make and model from the impression alone. This is genuinely useful. Knowing the offender wore a specific model of work boot in a size 11, with heavy outer-edge wear, narrows an investigation meaningfully and is a defensible class conclusion.
• Acquired (individual) characteristics: the randomly acquired marks a shoe picks up in service — cuts, scratches, gouges, holes worn through, a stone lodged in a tread channel, a chunk torn from a lug. Each such feature has a position, size, shape, and orientation. The premise of footwear identification is that the particular constellation of these accidental features on one shoe is effectively unique, and that if enough of them reproduce in the questioned impression and correspond in the known, the examiner can conclude that this shoe made this impression.

Here is where honesty matters. Footwear class evidence is on relatively solid ground — it is, at bottom, manufacturing information, and the inference "this impression is consistent with this model and size" is well supported. Footwear identification from acquired characteristics is more like the toolmark and firearms situation we come to in §16.4: it depends on the examiner's judgment that the corresponding random marks are numerous and specific enough to exclude every other shoe, and it has been subjected to far less rigorous error-rate study than DNA. It is more defensible than bite marks — the features are durable, measurable, and arise from a plausible random process — but the leap from "consistent with" to "made by this shoe and no other" is still a judgment call, not a calculation, and a careful examiner reports it as such.

⚖️ In the Courtroom The defensible footwear testimony ladder, from strongest support to weakest: "this impression could not have been made by the suspect's shoe" (exclusion — clean); "this impression is consistent with the class characteristics — the tread design and size — of the suspect's shoe, and of all other shoes of that design and size" (class association — well supported); "in addition to class agreement, I found corresponding randomly acquired features that, in my opinion, identify the suspect's shoe as the source" (identification — admitted in most courts, but it is an opinion about specificity, and cross-examination should press how many features, how distinctive, and against what known error rate). Watch for the examiner who skips from class agreement straight to "it's his shoe." Size and tread design are shared by thousands of people.

🧠 Cognitive-Bias Watch Footwear comparison is visual, subjective, and frequently done by an examiner who knows the suspect is in custody and which shoe the detective wants matched. That is precisely the recipe for contextual bias (Chapter 31): the examiner who expects a match counts ambiguous marks as corresponding and explains away differences. The safeguard is the same one most of pattern forensics still resists — document the questioned impression and reach preliminary conclusions before seeing the suspect's shoe, and have a second examiner verify blind. The number of acquired features needed to "identify" is a judgment, and judgment under expectation drifts.

16.3 Tire impressions

Everything true of footwear is true of tires, scaled up and with a few wrinkles of its own. A tire impression is the recorded tread detail left by a vehicle's tire — in mud at the end of a track, in snow, in a flower bed, or as a dust/residue transfer on a paved surface or across a victim's clothing in a vehicular case. Because vehicles place tires where a fleeing offender often cannot avoid soft ground, tire impressions can put a vehicle — though not, by themselves, a driver — at or near a scene.

The class-to-individual logic is identical. The class characteristics of a tire impression are its tread design (the molded pattern of ribs, lugs, and grooves), the tread width, and, when enough of the impression is present, the circumference and the noise treatment (the deliberately varied spacing of tread blocks that manufacturers use to reduce road hum — a pattern that repeats around the tire). From tread design and dimensions, an examiner can often determine the make, model, or at least the manufacturer and a set of possible vehicle types, the way footwear databases yield a make and model. The acquired characteristics are the cuts, stone-holds, nicks, and wear patterns a tire picks up over thousands of miles — and, as with shoes, a sufficiently distinctive constellation can in principle narrow toward an individual tire.

There are two tire-specific complications worth knowing:

1. A vehicle has multiple tires, and they may not match. Cars are often shod with tires of different brands, ages, and wear; the front pair wears differently from the rear. A complete impression sequence can therefore carry information about more than one tire, and an examiner records each. The good news is that this can increase discriminating power — matching the class characteristics of all four positions to one vehicle is more telling than matching one.
2. Tire impressions are long. A useful impression may run for several feet, and capturing it means photographing and casting in overlapping sections, then assembling them. The full circumference is rarely recorded in one impression, so the examiner works with whatever arc of the tread is present.

🔬 At the Bench Recovering a tire impression follows the footwear order — examination-quality photographs with a scale and oblique light first, then casting of three-dimensional impressions in dental stone, then lifting of two-dimensional ones. The wrinkle is length: because a single cast can only span so much, the examiner casts the impression in sections (a long impression in soil may take several adjacent pours) and labels their sequence so the lab can reconstruct the tread's repeating pattern. The comparison is then made against test impressions of the suspect vehicle's tires — typically made by inking or wetting the tire and rolling the vehicle across a long strip of paper or fabric to capture a full revolution.

⚠️ Junk-Science Alert Tire and footwear class evidence is sound; the danger, as always, is the overstated leap to a unique individual source. Be especially wary of "individual" tire identifications built on a short arc of tread with only one or two claimed acquired features. As with all impression identification, the field has far less published error-rate research than its courtroom confidence implies — a recurring finding of the 2009 NAS report across the pattern disciplines. "Consistent with a tire of this design and dimension" is honest and useful. "This impression was made by this specific tire, to the exclusion of all others" is a claim that needs many distinctive, well-reproduced acquired features and an examiner willing to state how few, or how many, they actually found.

Where does tire (and footwear) evidence sit on the validity spectrum? In the broad middle, and it splits by claim. The class-level inference — design, size/width, dimensions, hence make and model — is grounded in manufacturing fact and is among the more defensible comparison conclusions in the book. The individual-source inference shares the firearms/toolmark predicament we turn to next: plausible in principle, durable in its features, but under-validated as a quantified claim and dependent on examiner judgment. Reported honestly — class as class, individual only when the acquired detail genuinely earns it — impression evidence is a workhorse. Reported as routine certainty, it overstates.

🗂️ Read the Evidence — applied to the Cold Case (see §"The Case File" below for the full beat) The pollen evidence of Chapter 13 already told us a vehicle was at the Mill Creek cabin. Tire impressions, had the scene been worked properly, would have been the natural companion — a tread design and width to narrow which vehicle. Note the order of the inference: a tire impression places a vehicle, not a person, and certainly not a driver. The honest investigator keeps those three separate.

16.4 Toolmarks: pry bars, cut marks, and class-vs-individual claims

Now we move from impressions left by feet and tires to those left by tools, and the validity question sharpens. A toolmark is any impression, scratch, gouge, cut, or abrasion produced when a tool contacts a softer surface. The classic case is forced entry: a pry bar wedged into the gap between a door and its frame, then levered, leaves a compressed dent (the bar's profile pressed into the wood) and often a set of fine parallel scratches (the bar's edge dragged across the surface). Bolt cutters leave matched shear marks on a cut padlock or chain; a screwdriver used as a pry leaves the flat of its blade; a knife leaves striated cut marks on bone or cartilage. Toolmark examination asks: what kind of tool made this, and — the contested part — can we associate it with one specific tool?

Toolmarks come in two broad types, and the distinction governs what comparison can do:

• Impressed (or compression) marks, where the tool is pressed into the surface and leaves a negative of its shape — the dent of a hammer face, the profile of a pry bar's tip in soft wood. These carry mostly class information (the shape and dimensions of the tool's working surface) and sometimes coarse individual detail.
• Striated marks, where the tool slides along the surface and its edge cuts a series of fine parallel lines — the striations. Because the cutting edge of a tool carries a microscopic, irregular topography of nicks and burrs from manufacture and use, a striated mark is, in theory, the place individual identification might live: line up the questioned striations with test striations made by the suspect tool, and a correspondence of the fine lines is the toolmark examiner's analog of the firearms "match" you met in Chapter 15.

This is no accident of analogy: firearms identification is a special case of toolmark identification. A gun is a tool; a fired bullet and cartridge case carry striated and impressed marks from the barrel, breech, and firing pin (Chapter 15). The two disciplines share a professional body, a method, and — crucially — a validity verdict. So everything Chapter 15 said about the contested foundation of firearms "matching" applies here, and then some, because general toolmarks lack even the somewhat more constrained geometry of a rifled barrel.

The class–individual problem, and the subclass trap

Toolmark comparison is at its most defensible making class statements. Measure the width of the gouges in a doorframe and you can say the tool had a working edge of roughly that width; read the profile of an impressed mark and you can say the tool was a flat-bladed pry of a certain size — and therefore that the suspect's particular crowbar could have made it, along with every other crowbar of that width and shape. That is a real, useful, defensible conclusion: it supports forced entry and is consistent with a class of tool.

The trouble, exactly as in Chapter 1, is the leap to individual identification — "this mark was made by this tool and no other." Three problems dog that leap:

1. The subclass-characteristic trap. Recall the subclass characteristic from Chapter 1: a feature finer than the broad class but still shared by a subset of tools made by the same manufacturing process — tools cut or ground by the same machine in the same batch can carry similar fine marks. A striated toolmark "match" built on subclass marks will associate a questioned mark with a whole production run while looking like an individual identification. Distinguishing genuinely individual striations from subclass artifacts is the toolmark examiner's central skill, and it is a judgment, not a measurement.
2. Tools change, and marks vary with how the tool is used. Unlike a fired bullet (where the same barrel acts on each round in much the same way), a pry bar can be applied at any angle, with any force, on any material, and the resulting mark varies enormously. The examiner must make test marks that reproduce the questioned conditions — angle, pressure, substrate — and a poor reproduction yields a poor comparison.
3. No validated error rate for the individual claim. This is the heart of it. The 2016 PCAST report examined firearms/toolmark identification and concluded that its foundational validity as a method for individual-source identification had not been established by the kind of well-designed black-box studies that would measure how often examiners declare false matches. The discipline is admitted in most courts and is not without value, but the confident "this tool, to the exclusion of all others" outruns the published science.

🔬 At the Bench The toolmark comparison itself is done on a comparison microscope (Chapter 19) — two microscopes joined by an optical bridge so the questioned mark and a test mark sit side by side in one field of view, and the examiner rotates and translates them to bring striations into alignment. The procedure resembles the ACE-V framework you met for fingerprints (Chapter 14): analyze the questioned mark for suitability, compare it against test marks from the suspect tool, evaluate the agreement, and (ideally) have a second examiner verify. As with fingerprints, the method is only as objective as the examiner — there is no fixed numerical threshold for how many corresponding striations constitute a "match," and that absence is exactly what the validity critique targets.

⚖️ In the Courtroom — where toolmarks sit on the spectrum Place it honestly. Class-level toolmark testimony — "the marks are consistent with a flat pry bar approximately 19 mm wide; the suspect's crowbar could have made them" — is well supported and survives cross. Individual-source toolmark testimony — "the suspect's crowbar made these marks, to the exclusion of all other tools" — sits in the contested band of the validity spectrum (PCAST 2016: foundational validity not established). Increasingly, examiners and standards bodies favor more measured language ("consistent with," "cannot be excluded," or a stated degree of association) over categorical identification, and a sharp cross-examination asks for the error rate, the count of corresponding features, and whether subclass characteristics were ruled out — and how.

A note on relevance, threading back to Chapter 15. A toolmark, like any evidence, must be tied to the crime to matter. A pry-bar mark on a forced door supports forced entry — but forced entry by whom, and when? The mark dates and characterizes the act, not the actor. Keeping "what happened" separate from "who did it" is the discipline that the stray cartridge case taught us last chapter, and it applies in full to toolmarks: a class match to a pry bar is a fact about the door, not a name.

16.5 Bite-mark analysis: the rise and fall of a junk science

We come now to the discipline this chapter is named for, and the one the whole book uses as the far anchor of the validity spectrum opposite DNA. Bite-mark analysis is the claim that a forensic odontologist (a dentist with forensic training, Chapter 17) can examine a patterned injury on skin — a bruise or laceration left by human teeth — and determine which person's dentition made it, often to the point of identifying a single individual as the biter. It is, on the current scientific record, discredited: there is no validated method establishing that human dentition is unique in a forensically detectable way, that skin reliably records that uniqueness, or that examiners can reliably link a mark to a single biter. It has contributed to multiple wrongful convictions. And for roughly four decades it was confident, admitted, jury-persuading courtroom testimony. Understanding how that happened is the most important thing in this chapter, because the failure mode is general — it is how junk science gets into court whenever a method skips validation and goes straight to certainty.

The two assumptions, and why neither was ever validated

Bite-mark identification rests on a stack of premises. Stated plainly, they are:

1. The human dentition is unique — every person's arrangement of teeth (spacing, rotations, chips, missing teeth, the shape of the biting edges) is distinctive enough to tell people apart.
2. Skin (or another bitten substrate) faithfully records that uniqueness — a bite transfers enough of the dentition's distinctive features into the mark to support comparison.
3. A trained examiner can reliably compare a questioned bite mark to a suspect's dentition and identify the source.

Notice that even if premise 1 were true, the conclusion does not follow without 2 and 3 — and 2 and 3 are where the method collapses. Consider skin as a recording medium. Skin is elastic, mobile, and curved; it stretches and rebounds; the same teeth biting the same person twice can leave visibly different marks depending on the body location, the victim's position, movement during the bite, and the force. The injury then changes over time — a bruise spreads, swells, and fades, so the mark photographed at autopsy is not the mark made at the moment of biting. Skin is, in short, about the worst imaginable substrate for faithfully recording fine detail. It is nothing like the dental stone of a footwear cast or even the wood of a toolmark; it is a moving, healing, distorting surface.

And premise 1 — the much-repeated claim that dentition is unique — was asserted, not demonstrated. No one ever performed the study that would ground it: examine a large population, characterize the relevant dental features, and measure how often two people are indistinguishable as expressed in a bite mark on skin (not on a wax wafer, but on skin, healing, photographed days later). The "uniqueness of the dentition" is the bite-mark analog of the "uniqueness of the fingerprint" — possibly true at the level of the source object, but never validated as detectable and reliable in the trace actually recovered. As you learned in Chapter 1, uniqueness of the source is not the same as the ability to individualize from the impression, and bite marks fail the second test catastrophically.

⚠️ Junk-Science Alert This is the textbook case of the warning from Chapter 1: be most suspicious of any method whose central claim is "I can match this mark to one unique source by visual comparison" and which lacks well-designed studies measuring how often that matching is wrong. Bite-mark identification is that warning made flesh. Its practitioners projected confident, sometimes near-certain testimony ("the teeth that made this mark are the defendant's") onto a foundation that had never measured its own error rate — and when researchers finally tested examiners, the results were damning, as we describe below.

The rise: how it got into court

Bite-mark testimony entered American courts in the 1970s and gained a foothold through a small number of admitting decisions and the energetic advocacy of a handful of forensic dentists. Once a court admitted it, later courts cited those decisions; once it was "accepted," the Frye "general acceptance" standard (Chapter 5) helped it stay — accepted by whom? By the small community of forensic odontologists who practiced it, the same circular validation (a discipline admitted because its own practitioners vouch for it) that the 2009 NAS report identified across the pattern disciplines. A professional body, the American Board of Forensic Odontology (ABFO), developed guidelines and a scoring vocabulary, which lent the practice the appearance of standardization without ever supplying the missing validation. Examiners testified in terms of escalating confidence, sometimes using phrases that conveyed near-certainty to juries. The method had the trappings of science — a board, guidelines, credentialed experts, overlay photographs, courtroom exhibits — and almost none of its substance.

The fall: the reckoning

The collapse came from two directions at once, and they are the two engines of reform that recur throughout this book. The first was DNA exoneration (Chapter 6). As post-conviction DNA testing freed innocent people, several of the underlying convictions turned out to have rested on bite-mark testimony — the bite "matched" the defendant, but the DNA proved the defendant was not the source of the crime-scene biological evidence, and sometimes identified the true perpetrator. The method had not merely been weak; it had been wrong, with names and prison years attached. We meet two of those names in §16.6.

The second was formal scientific review. When proficiency-style tests were finally run — giving examiners bite marks of known origin and measuring their performance — examiners disagreed with one another and with the ground truth at rates incompatible with the certainty they had projected in court. They could not even reliably agree on the threshold question of whether a given patterned injury was a human bite mark at all, let alone whose. The 2009 NAS report found the scientific basis for bite-mark identification wanting; the 2016 PCAST report, applying its foundational-validity test, concluded that bite-mark analysis fell far short — there were no studies establishing that examiners can accurately associate a bite mark with a specific person, and the available evidence suggested they cannot. By the time of these reviews, even some former practitioners and the discipline's own organizations were retreating from the strong identification claims.

🧠 Cognitive-Bias Watch Bite-mark analysis is a cognitive-bias case study layered on top of a validity failure. The examiner is typically shown the suspect's dental models and asked, in effect, "does this bite match this person?" — a question that invites confirmation. With a distorted, ambiguous mark on skin and a known suspect to compare against, the examiner's expectation does much of the matching, finding correspondence in noise. This is contextual bias (Chapter 31) operating on an unvalidated method — the worst of both worlds. A weak method plus a biased procedure is precisely the combination that convicts the innocent, and bite marks delivered both.

Where it actually has value — and where it does not

We must be precise, because the same dentists who do bite marks do something genuinely valuable. Forensic odontology as a field of human identification — comparing a deceased person's teeth to their antemortem dental records to confirm who they are (Chapter 17) — is reliable and legitimate. Teeth are durable, survive fire and decomposition, and the comparison is to that person's own documented records, a fundamentally different and far better-grounded task than guessing whose teeth made a bruise. The next chapter, on odontology, deliberately holds these two faces side by side: valid dental identification of the dead, invalid bite-mark identification of the living biter. Same specialty; opposite ends of the validity spectrum. The lesson generalizes one last time: a discipline is not valid or invalid as a whole — a specific claim is validated or it is not, and you must always ask which claim is being made.

⚖️ In the Courtroom — the current status Bite-mark testimony is increasingly excluded or sharply restricted. A number of courts and scientific bodies have moved against it; some jurisdictions now bar testimony that a particular person made a particular bite. Where any bite-mark testimony is still admitted, the honest scope is narrow — at most that an injury is consistent with a human bite, and possibly that a suspect cannot be excluded — and never that "these teeth and no others made this mark." On cross-examination, the questions write themselves: What is the measured error rate for associating a bite mark with an individual? Can you reliably distinguish a human bite from another patterned injury? Has any study shown examiners can do what you just told the jury you did? For bite-mark identification, the honest answers gut the testimony.

16.6 The exonerations bite marks caused (Krone, Brown)

The case against bite-mark analysis is not abstract. It is written in the years that innocent people spent in prison, and in two cases the science of this book reads as object lessons. Both are real, both are matters of public record, and both end the same way: a confident bite-mark "match" to an innocent man, overturned by DNA. Treat them clinically — the human cost is the point, not the spectacle.

Ray Krone — "the Snaggletooth Killer"

Ray Krone was convicted in Arizona in connection with the 1991 murder of a woman in a bar where he was an acquaintance. A central pillar of the prosecution was bite-mark evidence: a forensic odontologist testified that bite marks on the victim matched Krone's distinctive dentition — his crooked teeth earned him the tabloid label "the Snaggletooth Killer." Krone was convicted and sentenced to death. On appeal his conviction was overturned on an unrelated procedural issue, he was retried — and bite-mark testimony again featured — and again convicted, this time receiving a life sentence. He maintained his innocence throughout.

In 2002, DNA testing of biological evidence from the crime excluded Krone and matched another man, already in prison for an unrelated offense, who had no connection to Krone. After more than a decade of imprisonment, much of it under a death sentence, Ray Krone was exonerated and released. He is often cited as among the people freed from death row by post-conviction DNA testing in the United States.

🔬 Read the Evidence

text FIGURE 16.2 — "Two methods, one defendant, opposite answers" [after Ray Krone, public record] THE ITEM The bite-mark testimony that helped convict Ray Krone (an odontologist's opinion that injuries on the victim matched Krone's distinctive teeth) set against the later DNA result from crime-scene biological evidence. THE CONTEXT Two comparison methods applied to the same case at different times: an examiner's visual bite-mark comparison at trial; nuclear DNA typing years later. WHAT IT SHOWS The DNA excluded Krone and matched a different, identified man. The bite-mark "match" had pointed confidently at the wrong person. WHAT IT DOESN'T The DNA does not retroactively make the bite-mark method "almost right" — it exposes that the method's confident identification carried no real reliability. THE INFERENCE A method at the DISCREDITED end of the validity spectrum (bite marks) was overruled by one at the STRONG end (DNA). The bite "match" was not weak evidence pointing the right way; it was an unvalidated method pointing the wrong way. THE LESSON Confidence is not validity. The most persuasive courtroom exhibit — an expert with two images and a pointer — can be the most wrong. Ask for the error rate before you ask what the expert concludes.

Roy Brown — the bite that excluded its own defendant

Roy Brown was convicted in New York in connection with a 1991 murder in which the victim suffered multiple bite wounds. Bite-mark testimony linked the marks to Brown — but the case carried an extraordinary detail that, read honestly, should have excluded him at trial. The bite marks reportedly showed features inconsistent with Brown's dentition; by one account the marks displayed more upper teeth than Brown possessed, a discrepancy the trial testimony minimized rather than treated as the exclusion it arguably was. Brown was convicted and spent roughly fifteen years in prison.

Brown himself, from prison, pursued another suspect through public-records requests. After his release was secured, DNA testing of saliva on the victim excluded Brown and implicated the other man. Roy Brown was exonerated. His case is doubly instructive: not only did bite-mark "identification" point at an innocent man, but the very evidence used to convict him contained, on its face, a difference — the kind of unexplained discrepancy that, in any disciplined comparison, is grounds for exclusion. The method was so unmoored from the exclusion-over-proof logic of this book that an apparent mismatch was spun into a match.

⚖️ In the Courtroom Roy Brown's case is the courtroom embodiment of §16.1's hierarchy of conclusions. A genuine, unexplained difference between the questioned mark and the suspect should drive toward exclusion — that is the most defensible thing impression comparison does. When an examiner instead explains away a difference to preserve a match the prosecution wants, the discipline has inverted its own logic. The first question on cross in any comparison case is therefore not "what agrees?" but "what differs, and why isn't that an exclusion?"

What the two cases teach — all four themes at once

These exonerations are not just sad stories; they are the book's four themes converging, which is why this chapter holds them up:

1. Exclusion over proof. In both cases the cleanest, truest forensic act came late — DNA excluding the convicted man. Brown's case shows the tragedy of an exclusion ignored: a difference was available at trial and was overridden. Forensic science's surest voice (the "not this person") was present and silenced.
2. Not all methods are equally valid. Bite marks (discredited) versus DNA (strong) is the validity spectrum at its widest, applied to the same defendants. The method that convicted them sat at the bottom; the method that freed them sits at the top.
3. Cognitive bias is the chief threat. Examiners shown a suspect and asked to confirm a match found one in distorted, ambiguous marks — confirmation and contextual bias operating on a method with no error rate to discipline it.
4. The CSI effect cuts both ways. Juries, primed to trust a credentialed expert with side-by-side images, heard certainty where there was none, and over-trusted weak evidence — exactly the second, less-discussed direction of the CSI effect from Chapter 1. The confident bite-mark exhibit is the over-trust failure in its purest form.

The honest summary is stark and worth stating without hedging: bite-mark identification is the clearest example in the entire field of a method that was admitted, believed, and wrong, and the cases of Ray Krone and Roy Brown are part of why it is now, rightly, in retreat.

🗂️ The Case File

Carrow County — the door, the soil, and the mark that was never there. Return to the Mill Creek cabin. Two impression findings, properly worked, would belong in the file this chapter; one telling absence completes the lesson.

The door (toolmarks). Recall from the Chapter 3 inventory that the cabin's door carried pry marks around the lock area. Examined as toolmarks, those gouges and the compressed profile in the doorframe are measured and characterized: their width and shape are consistent with a flat pry bar — a class of tool — supporting forced entry. This is a genuine, defensible conclusion at the class level, and you should state it at exactly that strength: the marks are consistent with a pry bar of a certain width and profile; such a tool could have made them. What the toolmarks do not establish, on their own, is which pry bar, or whose, or when. There is no validated method here for individualizing a general toolmark to one tool to the exclusion of all others (PCAST 2016 on firearms/toolmarks), and even if there were, the mark would date the act of forcing the door, not name the person who held the bar. Honest status: forced entry is supported — a significant finding, because forced entry is hard to square with the original "accidental fire, a man who didn't get out" story. A man dead in a fire does not, by himself, force his own door from outside.

The soil (footwear). A partial shoe impression is recorded in the soft soil outside that door. Worked correctly — photographed with a scale under oblique light, then cast in dental stone — it yields class characteristics: a tread design and an approximate size. That narrows, modestly, the population of footwear that could have left it, and it can be compared against the shoes of persons of interest as they are identified. State it honestly: an impression consistent with a tread design and size is class evidence; it is not, from a partial print, an identification of one shoe. (Pair this, eventually, with the pollen of Chapter 13 and the soil comparison of Chapter 24: independent, modest associations that converge are the whole strategy of the cold case.)

The mark that isn't there (bite marks). And note what the file does not contain: there are no bite marks in this case, and that absence is itself this chapter's lesson. Had a patterned injury been present and had an over-eager examiner "matched" it to a suspect, you now know exactly how little that would be worth and exactly how it could mislead — Krone and Brown are the warning. The most valuable thing bite-mark analysis contributes to the Mill Creek file is the discipline of not using it: a reminder that a confident comparison from an unvalidated method is worse than no comparison at all.

Running status after Chapter 16: Forced entry supported (class-level toolmark); a footwear impression logged as class evidence pending comparison; no bite-mark evidence, and none would be trusted if there were. The "accidental fire" frame, already cracked by the pathology to come, is further strained: someone, it appears, forced their way in. No person is yet placed at the door — that work is still ahead.

Conclusion

Impression and toolmark evidence is the validity spectrum compressed into a single chapter. At one end, footwear and tire class evidence — tread design, size, dimensions — is grounded in manufacturing fact and is among the more defensible comparison conclusions in forensic science; recovered properly (photograph first, then cast) and reported as class evidence, it is a workhorse that narrows investigations honestly. In the middle, toolmark and footwear individual identification shares the firearms predicament of Chapter 15: plausible in principle, dependent on examiner judgment, threatened by the subclass trap, and under-validated as a quantified claim — defensible as "consistent with," contested as "to the exclusion of all others" (PCAST 2016). And at the far end sits bite-mark analysis, the field's most thoroughly discredited discipline, which projected near-certainty onto skin — an elastic, healing, distorting surface — from a foundation that never measured its own error rate, and which helped convict Ray Krone, Roy Brown, and others later proven innocent by DNA.

The transferable lesson is the one the whole book keeps teaching: a discipline is not valid or invalid as a whole; a specific claim is validated or it is not, and you must always ask which claim is being made and whether anyone has measured how often it fails. The same forensic odontologist who cannot reliably tell whose teeth made a bruise can reliably identify a body from its dental records — which is exactly where the next chapter goes. Chapter 17 takes up forensic odontology as human identification: the valid, durable use of teeth to name the dead, deliberately set beside the invalid bite-mark use we have just buried. And in the cold case, it is dental identification that will confirm whose body lay in the Mill Creek cabin — closing a door this chapter's forced one helped open.

Key Terms

• Toolmark — any impression, scratch, gouge, cut, or abrasion produced when a harder tool contacts a softer surface; analyzed as class (the tool's general shape/size) and, contestably, individual (fine striations) evidence.
• Impression evidence — physical evidence in which the shape of one object is recorded on the surface of another (footwear, tire, tool, bite), governed by the class-vs-individual comparison logic.
• Footwear impression — the recorded outline and tread detail left by the underside of a shoe; carries class characteristics (tread design, size, wear) and acquired/individual characteristics (cuts, gouges, lodged debris).
• Tire impression — the recorded tread detail left by a vehicle's tire; class characteristics (tread design, width, dimensions, noise treatment) can yield make/model, placing a vehicle (not a driver) at a scene.
• Cast — a three-dimensional reproduction of an impression, made by pouring a casting medium (typically dental stone) into a three-dimensional impression to create a positive replica for comparison.
• Bite-mark analysis (debunked) — the discredited claim that an examiner can identify the person who made a patterned bite injury, especially on skin; lacks validated foundation, fails error-rate testing, and has caused multiple wrongful convictions.

Spaced Review

1. A footwear impression yields a tread design and a size, but the partial print shows no distinctive cuts or gouges. What is the strongest honest conclusion an examiner can offer, and why is "it's the suspect's shoe" an overstatement? (§16.2)
2. Explain the subclass-characteristic trap in toolmark comparison, and why it can make a production-run match masquerade as an individual identification. (§16.4; recall the term from Chapter 1.)
3. Bite-mark identification assumes the human dentition is unique. Even granting that assumption, name the two further premises the method needs — and explain why skin makes one of them fail. (§16.5)
4. Validity-spectrum question: Where do (a) footwear class evidence and (b) bite-mark identification sit on the NAS 2009 / PCAST 2016 validity spectrum, and what single question best distinguishes them? (§16.1, §16.5)
5. In Chapter 1 you learned that forensic science excludes more reliably than it proves. Using the Roy Brown case, explain how a comparison discipline can fail by ignoring an exclusion. (§16.6; Chapter 1, §1.6)