In 2003, British police arrested bartender Peter Hamkin on suspicion of murdering a woman in Italy the year before. Now, bartenders know a lot of tricks, but killing complete strangers in countries they've never visited is not among them. Was this another case of eyewitness accounts gone wrong? No, although officials said he matched the description of the assailant. This time, the culprit was that unassailable mainstay of crime procedurals, DNA [source: Geddes].
DNA database matching actually compares only a selection of subsites on the strand known as loci. American labs use 13 loci, while in the UK 10 is the magic number, and suspects need not match all of them. If that sounds scary, consider this: Because countries don't always use the same areas, a proposed pan-European database would require a mere six loci for a match [source: Geddes].
This number once seemed like plenty — experts placed the chances of a false match among unrelated people at 1 in 113 billion. But in 2008 Arizona state crime lab analyst Kathryn Troyer found dozens of such matches when using the nine-loci standard common at the time, which suggests the subject might benefit from some reconsideration. It also sent lawyers nationwide on a genetic fishing expedition [source: Felch and Dolan].