In 1903, a new prisoner named Will West arrived at the United States Penitentiary at Leavenworth, Kansas, and was brought in to be measured and photographed, as every new prisoner was. The clerk looked him over and said he had been there before. No, sir, West answered. The clerk ran his calipers and tape over him anyway, reducing him to a set of numbers the way Alphonse Bertillon's French system required, then carried the figures to the files to see what they matched.
The match came back at once, and it was wrong. The card showed a face that looked like West's own, a name close enough to be his own, and eleven body measurements that lined up with the ones just taken almost digit for digit; none of the eleven differed by more than a centimeter. The name on the card was William West, already a prisoner in that same institution, serving a life sentence for murder since September 9, 1901. He and the new arrival were not known to be related. Shown the card, West laughed and said it was his own picture, though he had never set foot in Leavenworth before that week.
Bertillon's system rested on a simple bet: that enough independent measurements of a body, taken together, could not plausibly land on the same numbers twice. That bet held up well enough for police and prisons across several countries to file criminals by the centimeter rather than by name or photograph, both easy to fake or forget. The two West men broke the bet without anyone cheating. When their cards were later reproduced side by side in a textbook on personal identification, the only honest comment left for the authors to make was that the numbers could have come from one man, measured twice, by two different clerks.
Fingerprints showed what the measurements could not. Will West's left index finger was classified as a whorl, with eleven ridges counted between its core and delta; William West's was classified as a loop, with eighteen. The two classifications had nothing in common, the kind of difference fingerprint examiners expect between any two strangers off the street. The Federal Bureau of Investigation's own account of the case, published decades later, said plainly what it had cost the older system: one case had exposed the limits of photographs, measurements, and names at once, while fingerprinting was the method that had not failed.
Bertillon's method, introduced in Paris in December 1882 and in use for about twenty years by the time Will West reached Leavenworth, was a serious, internationally adopted attempt to make a stranger's identity a checkable fact instead of a guess. Leavenworth's clerk did not fail by being careless; he measured West by the book, and the file handed him back a different man anyway. Bertillon himself had written that a few millimeters of difference between two signalments proved nothing on their own, since no two measurements of even the same person ever come out identical; what his system assumed was that enough measurements taken together, on two different bodies, would not plausibly add up to a false match. Across eleven measurements of two unrelated men, they did.
Fingerprints did not replace Bertillon's calipers because they were a newer idea. They replaced them because a fingertip holds more ways to differ than a tape measure can find across an entire body, and a system for telling people apart is only as good as the rarest thing it actually measures.