Bitemarks

Last time we talked about bitemarks, it was about how it isn’t supported as a method and wondering why professional organizations still support it, even if quietly. The NIST study did a very good take-down of bitemarks but another contender has been published in the Journal of Law and the Biosciences, titled, “Forensic bitemark identification: weak foundations, exaggerated claims.” The article has an impressive list of authors (38!), including legal experts, scientists, statisticians, medical examiners, and odontologists. The authors provide excellent context as to how we got here in the first place with odontology and give some insightful commentary on other methods and disciplines in forensic science*. They note that it has taken more than three decades to even start to undo this “massively unsupported field,” which is as much an indictment of our courts as it is of any science*.

I heartily suggest you read the paper in its entirety (it’s free)¹.

How did we get here?

Forensic dentists had been content identifying the deceased through dentition and pre-mortem records, like x-rays. They knew using bitemarks on substrates like skin was a difficult, at best, proposition:

Until 1974, the discipline refrained from trying to identify the source of a bite mark left in skin because the differences between identifying victims of mass disasters and identifying the source of a crime scene bite mark seemed to them prohibitively daunting:

The two tasks differ in important ways. In the disaster situation, there is a finite number of candidates to identify, and full dentition often is available from the victims as well as from the dental charts. In forensic bitemark cases, the number of potential suspects is huge, the bitemarks include only a limited portion of the dentition, and flesh is a far less clear medium than having the teeth (of the disaster victim) themselves.

Thus, crime scene bite marks contain only a small fraction of the information available from the full dentition of mass disaster victims, and the limited dental information that is available is neither clear (flesh is far from an ideal medium for recording bite marks) nor dependably accurate (flesh is elastic and subject to distortion at the time of and after receiving the bite).

But in the California case of People v. Marx (1975), forensic dentists were asked to determine if a bite mark on a murder victim's nose matched the defendant's teeth. This was unusual, as bite marks weren't generally trusted to accurately identify suspects. However, this case was different: the bite mark was very clear and the suspect's teeth were highly unusual. The dentists testified that the bite mark was a match, and the defendant was convicted. The defense argued that this type of evidence was new, and that the dentists' were implying statistics (a la Collins). The court disagreed, stating that the methods used—dentition, x-rays, models, photography— were established techniques and that the dentists didn't explicitly use math, so their conclusions were admissible.

The most sensible, and charitable, reading of Marx would be that the court understood, along with the forensic dentists, that the circumstances of the injury presented an unusually stable bite mark of an apparently very unusual set of teeth. In short, the offer and the admission in Marx constituted a rare exception to the general rule (among forensic dentists) that bite marks were a poor basis for trying to compare patterns…What had been an exception to the rule magically became the rule, not only for courts but for forensic dentists as well. But, ironically, rather than forensic dentists convincing courts that their field could accurately identify the sources of bite marks, the courts convinced forensic dentists that they could do]what until then they doubted they could do. [emphasis added]

As much as one may rail about bad forensic science*, it must be remembered that some degree of fault lies on the attorneys and judges who promulgate it. Attorney’s like agreeable witnesses because “you know what the son-of-a-bitch is going to say.” Forensic science* is the subordinate player in the criminal justice system, so we either go along or we don’t go to court.

Fidelity

As with all evidence, the faithfulness or fidelity of the signal from the source to the receiver is of paramount importance. It is no less so with bitemarks:

…the faithfulness of the transfer from the original to the receiving surface, and the ability of the receiving surface to retain the impression unchanged, are essential to the probativeness of the comparison of the mark on the receiving surface to a suspected source.

Without a clear signal, the transferred evidence is just so much noise. Sometimes this is due to the inherent lack of information in the evidence (like with white cotton fibers), in the transmission (blurred, distorted fingerprints), or the substrate (gravel receives shoe prints poorly). The questioned item demands closer attention, because unless we find sufficient information in it to classify it, we should not go to the comparison stage.

“Match”

There’s that word again. It’s as bad as “identify,” in some ways. The authors of this paper bring up a point about “matches” and interpretation I’ve made before, usually after a martini:

That is not to say it must be done just as DNA typing does it. But without some methods for evaluating the meaning of a suspected source having similar appearance to the crime scene evidence, a factfinder has no way to gauge how probative that fact is, and might be misled by testimony saying only that a suspected source has been judged to ‘match’ the crime scene mark—in whatever terms that fact might be expressed.

Except for DNA (ahem), we can’t say the chance of finding another source entity like this one is X in Y or that this entity occurs this often at random or whatever. Interpreting evidence is inherently probabilistic but we’re bereft of useful numbers. “Could have come from,” “failed to exclude,” and other mushy ways of trying to express statistics without using statistics…because we don’t have any…are the mainstay of many forensic reports. You want to use likelihood ratios? Nice. What number do you put in for the prevalence of polyester fibers, 15nm in diameter with a trilobal cross-section with a modification ratio of 2.2 dyed with disperse dyes to this specific color?

Take your time. I’ll wait. No hurry. I’ll just make another martini while you look that up. By the way, would knowing the fiber’s birefringence and subpolymer type help? Whether it's a staple or filament fiber? I could analyze the dyes (there’s more than one in there, bucko) and give you those too, if that would help. Uh huh. Thought so.

Want to go into forensic academia? Learn statistics, coding, AI and save our butts. Please. I’m too old for that.

Anyhoo, read that article.

1. If you don’t have access to anything I link in these posts (and I try to use open access as much as possible), contact Florida International University’s Forensic Research Library. It’s free! [Disclaimer: My day job is at FIU.]