Chapter 6 — Reading Real Traces

Chapter 6 — Reading Real Traces

In two borrowed attic rooms above the Lyon courthouse, in 1910, a Frenchman named Edmond Locard set up the first real forensic laboratory in the world and began, in earnest, the work that Sherlock Holmes only pantomimed. He had read the stories. He admired them; he cited Doyle by name. And then he did the thing Doyle never had to: he tried to make it survive contact with reality — which meant, above all, making it survive being wrong.

Hold that date in your mind next to the other one. Holmes first appeared in 1887. Locard's lab opened in 1910. Francis Galton did not publish the statistics that made a fingerprint mean something until 1892. Hans Gross did not coin the word "criminalistics" until 1893. Scotland Yard's Fingerprint Bureau opened in 1901. In other words: nearly the entire real science of physical evidence was built after Sherlock Holmes, not before him. Doyle did not dramatise an existing science for a lay audience. He dramatised a hope — a single brilliant man's hunch — and then the world spent the next thirty years doing the laborious, error-haunted, unglamorous work required to earn the certainty Holmes had been handing out, gratis, since 1887. The fiction came first. The science ran to catch up. And catching up meant learning the one lesson the fiction had skipped: how to be humble about your own mistakes.

Every contact leaves a trace — sometimes

Locard's name survives in a single sentence, the closest thing forensic science has to a creed: "Every contact leaves a trace." Touch anything and you take some of it with you and leave some of yourself behind — fibres, hairs, skin, soil, the microscopic exchange that ties a person to a place. It is a genuinely profound idea, and it is the deep truth of which Holmes's cigar-ash theatrics were the cartoon. It means the world really is written on, constantly, by everyone who moves through it. The physical record is real.

But notice how the actual scientist states it, and notice what the honest version costs. "Every contact leaves a trace" is not a promise that the trace can be found. Nor that it can be read. Nor that, once read, it points to one and only one person. Locard's own writing was more hedged than the ringing motto suggests — the crisp slogan is a later polishing of a more careful, more qualified claim. The principle guarantees that evidence exists; it guarantees nothing about our ability to recover it, interpret it, or keep ourselves from reading into it the story we already wanted to find. That gap — between "a trace exists" and "and therefore I know who did it" — is the entire discipline of forensic science, and for most of the twentieth century, we were nowhere near as good at closing it as we told ourselves, and told juries, we were.

The letter that graded the whole profession — and most of it failed

For a hundred years, forensic scientists testified with Holmesian confidence. This hair matches the defendant's. This bite mark identifies him. These two bullets were fired by the same gun, to the exclusion of all others. Juries heard the certainty of the great detective, in a white coat, under oath. And then, in 2009, the profession received the most brutal report card in its history.

The United States Congress had asked the National Academy of Sciences to survey the state of forensic science. The resulting report, Strengthening Forensic Science in the United States, landed like a demolition charge. Its central finding, stated with devastating plainness: "With the exception of nuclear DNA analysis... no forensic method has been rigorously shown to have the capacity to consistently, and with a high degree of certainty, demonstrate a connection between evidence and a specific individual or source." Read that again with the weight it deserves. With one exception, none of it — not bite marks, not hair, not fibres, not tool marks, not handwriting, not the confident matches that had helped convict people for decades — had been scientifically validated to do what its practitioners routinely swore in court that it could do.

A second landmark report, from the President's Council of Advisors on Science and Technology in 2016, took the pattern-comparison disciplines one at a time and sorted them by whether they had any demonstrated "foundational validity" — any actual studies measuring how often they get it right and wrong. The results split the field in two. On the validated side: single-source DNA analysis, and, with real caveats about examiner error, latent fingerprints. On the not-established side, a roll-call of techniques that had been sending people to prison for years — bite-mark analysis, footwear matching, microscopic hair comparison, firearms and tool-mark identification, bloodstain-pattern interpretation. The emperor was found to be, if not entirely naked, then wearing far less than his courtroom testimony had claimed.

The human cost of the gap is not abstract. When the FBI, the Department of Justice, and the Innocence Project jointly reviewed cases in which FBI examiners had given microscopic hair-comparison testimony, they found erroneous statements favouring the prosecution in more than ninety-five percent of the cases examined — hundreds of them, some capital. Cameron Todd Willingham was executed by the state of Texas in 2004 for setting the fire that killed his three children; the arson "science" that convicted him — the char patterns and crazed glass that "proved" an accelerant — was subsequently and comprehensively discredited as folklore. Bite-mark evidence, once presented to juries with total assurance, has been linked to a long series of DNA exonerations. This is what a discipline looks like when it inherits Holmes's confidence without Holmes's author to guarantee the confidence is deserved. The men in the ditch from Chapter 2 turn out to have names, and case numbers, and in Willingham's case a grave.

The fingerprint that framed an innocent man

Even the good techniques bleed at the edges, and honesty about the edges is exactly what separates real forensic science from the fictional kind. Take fingerprints — one of the disciplines that came through the audits comparatively well. For a century, examiners testified to a zero-percent error rate; a fingerprint match was presented as absolute, individual, infallible.

Then came Brandon Mayfield. In 2004, after the Madrid train bombings, the FBI matched a fingerprint from a bag of detonators to Mayfield, an Oregon lawyer — a match three separate FBI examiners called one hundred percent certain. Mayfield had not been to Spain in years. He was a Muslim convert, which did not help the Bureau's objectivity. He was arrested and held. And he was completely innocent: the print belonged to an Algerian national, as the Spanish police had been trying to tell the FBI the whole time. The "certain" match was wrong — undone in part by exactly the confirmation bias of Chapter 4, the examiners seeing what the context had primed them to see. The myth of the infallible fingerprint died in that cell, and its death made the science better, because a discipline that admits its error rate can measure it, control it, and warn the jury about it. A discipline that claims infallibility can only wait to be caught.

The humbling that made it powerful

Here is the turn, and it is the opposite of what a Holmes fan expects. You might think this chapter is bad news for the resurrection — that if forensic science is this fallible, the dream is dead after all. Precisely backwards. The 2009 and 2016 reports are not the humiliation of forensic science. They are the moment it became a science, because becoming a science is exactly the act of turning around and interrogating your own reliability, counting your own errors, publishing your own failure rates. Holmes never once did this. He could not; his author forbade it. Real forensic scientists did it to themselves, in public, under congressional letterhead — and the disciplines that survived the audit, chiefly DNA, are now more genuinely powerful than anything in the canon. DNA evidence has both convicted the guilty and, through the Innocence Project and its kin, walked hundreds of wrongly convicted people out of prison, including many put there by the junk sciences the reports exposed. That is the true heir to Locard's principle: not a man announcing a suspect's biography from his watch-chain, but a discipline humble enough to say "this technique is right this often, and wrong this often, and here is how we know" — and powerful because it can say it.

The trace, then, is real. The reading of it is hard, humbling, and worth doing right. But there is one kind of trace we have not touched, the one Holmes prized above all others and the one modern science is most sceptical of: not the fibre or the footprint, but the tell — the flicker of the face, the tremor in the voice, the story that doesn't add up. Can you really read a person? Can anyone? The answer is stranger and funnier than either Holmes or his debunkers would have you believe, and it is where we turn next.