Fingerprinting Gone Wrong: The Madrid Bombings

Welcome to CS-Science!

​

Fingerprints! We all have them and they are one of the oldest forms of forensic science.

​

And it’s pretty straight forward, you dust for fingerprints at a crime scene run the prints against the database – and voila, it’s a match! Right? We’ve all seen it on TV shows like this one.

​

Just pop it in the computer, some whurs and beeps and bam – it’s a match. Easy peasy.

​

But in 2004, an American citizen was arrested and imprisoned based on fingerprint evidence, only to be later released. So what went wrong?

​

In 2004 a bomb ripped through a train in Madrid, killing 193 people and injuring around 2000 more. Forensic investigators found fingerprints on a bag at the scene that contained the detonators. They dusted them, lifted them and sent a copy to the FBI in the USA.

​

The FBI ran the fingerprint through their databases, and… It’s a Match!

​

The fingerprint matched that of Brandon Mayfield, an American lawyer who had recently converted to Islam.  He was arrested and taken into custody. There was only one problem, he hadn’t been in Spain – in fact he hadn’t left the country at all, for years. So how did this happen?

​

Spanish police soon identified another match for the fingerprint, an Algerian national who had been in Spain, and Brandon Mayfield was released from prison. So what went wrong?

​

Fingerprint analysis works on the assumption that everyone’s fingerprints are unique, but how do forensic scientists measure this ‘uniqueness’?

​

Forensic scientists examine features in fingerprints called whorls, loops and arches. But matches are not made on the whole fingerprint, but by comparing individual features on the prints.  So how many of these features does a fingerprint analyst have to look at before they can decide that it’s a match? The answer is ‘it depends’ – in the USA there is no rule about the minimum number of matching features needed to declare a match.

​

But isn’t it all done by computers these days anyway? Yes and no. In the case of Brandon Mayfield, the FBI’s computer database compared the print to 45 million people in its database, and returned 15 potential matches. But these matches still had to be examined by forensic scientists, yes, that’s right, actual people, to determine the final match.

​

And it has been found that bias and context can influence the result. Researchers from University College London gave a crime scene print and a suspect print to five analysts, and they all correctly identified that the prints matched. Five years later they were sneakily given THE SAME pair of fingerprints but this time they were told that it was the print found on the bag at the Madrid bombings, and the print that FBI matched to it, that is setting the expectation that it would NOT be a match.

​

Only one forensic scientist maintained that the prints were a match, three changed their mind and now said they weren’t a match and one said they were undecided!

​

So, in the case of the Madrid bombings, were the FBI fingerprint analysts influenced by knowing the suspect’s background and religion? An investigation in the botch up claimed no, they did not know these things when making the initial match, but admitted that there were questions about why it took them so long to revisit their analysis and release Mayfield when the Spanish police had already arrested another suspect, who was actually in Spain.

​

So the truth of fingerprint analysis is that unlike TV shows where it is instantaneous and the computer has the final say, people are still involved in the analysis and this means that things can sometimes go wrong.

​

That’s all from me today! Be sure to check back soon for the next video where we put the science into CSI---ence.