The "Phantom" of Heilbronn and Negative Controls (Or the Lack Thereof)

Mar 27 2009 Published by under Science

It looks like there are quite a few police officers in Germany with egg on their face right now. They spent several years, thousands of man-hours, and over $14 million trying to track down a criminal mastermind, only to discover that they were chasing lab contamination. Sadly, I am serious about this. From a 2008 Telegraph article:

Police in Germany have stepped up the hunt for a serial killer nicknamed "the woman without a face".

The mystery woman has been linked by DNA to six murders and a string of thefts in a 15-year spree in three countries. Her latest victims may be three second-hand car dealers shot execution-style.

The inquiry is based entirely on DNA found in smudges of sweat and nearly invisible flakes of shed skin at the crime scenes.

The malevolent and depraved fiend had, by that time, engaged in a crime wave that crossed international borders. Evidence of her heinous acts was found throughout much of southern Germany, as well as in neighboring parts of Austria and France. Oddly, though, it appeared that the criminal might have had some sort of phobia or aversion to Bavaria - no trace of her was ever found there. (I'll get back to that in a bit.)

There were two major problems that the German police encountered along the way. One of the issues involved an over-reliance on DNA, even in the face of conflicting evidence. The other involved their own forensic procedures.

It's hard to look at some of the accounts of the criminal activity associated with the "phantom" culprit without wondering how the police could have managed to avoid concluding that they might just possibly be dealing with a contamination problem.

One clue that the police seem to have missed involved the mystery woman's accomplices:

Police revealed that other DNA traces were found at crime scenes indicating she sometimes operated in tandem with another. But no two crime scenes yielded the same DNA, indicating she picks up and discards helpers with the same casual abandon with which she kills.

Kurt Kletzer, a noted Viennese psychiatrist, says the Woman Without a Face is "intriguing and disturbing" in equal measure.

Intriguing indeed. Multiple accomplices, a massive reward, and not one of them steps forward to turn her in? I'm not an expert in the criminal mind, but that sounds unusual.

The police also seem to have failed to notice some other anomalous bits of data. For example, there were the traces that she left at a crime scene in 2005:

On May 2005, in the city of Worms, a local gypsy turned a gun on his brother. Police later found the phantom's DNA on one of the bullets.

More recently, there were signs that she turned up somewhere else where her involvement might not have been expected:

This year, police found a few cells of her skin after they stripped and analysed all the upholstery and carpets from the car of a man they are holding on suspicion of triple murder.

A former police informant, he was suspected of killing the three car dealers who had come from the Caucasian republic of Georgia to buy second hand German cars. Their bodies were dumped in a river at the end of January.

The informant denies the charge, saying another man, a Somali Islamist, also in police custody, was the killer. He also denies any knowledge of the phantom woman.

But her DNA has been found in the car, triggering a new theory that she was the executioner.

Police are compiling a thorough history of the car: who owned it, where it has been seen. They already know the three Georgians were driven in the car to Heppenheim, the town where they were killed.

Odder still, the car was owned by the police at the time of the killings.

The criminal investigation department bought it second-hand last year and loaned it to the informant when he was employed to brief police on the activities of his criminal associates.

Erwin Hetger, police chief of Baden-Wuerttemberg state, was jubilant at the find, calling it a "down payment" for police to solve the case of the woman without a face.

"We're closing in on her," he said.

Despite all of those signs, it wasn't until very recently that the police discovered and admitted that they'd been trying to find a phantom for all those years.

After spending millions of pounds and hundreds of thousands of police man hours searching for "The Woman Without a Face" - who reportedly struck across Europe and had been linked to six murders - German police have revealed that they have probably been following the DNA of a factory worker who handled the sterile cotton swabs used by police forces across Europe.

It seems fairly clear that the police had become so committed to the image they'd built around this phantom woman that the idea that she might not really exist simply did not occur to them. She had become, in their minds, as real as anyone they knew - if not moreso. They'd spent time trying to get into her mind, and "she" had gotten into theirs. I would be surprised if this case isn't written about in psychological journals and textbooks in years to come.

Hopefully, it will also be written about in criminal justice and forensics texts, and not merely as a warning to avoid becoming overly-committed to any one line of evidence. This entire embarrassing debacle could, and should, have been avoided. If the police forensic techs had been using good lab procedure, I would not be writing this now.

Every DNA lab in every university that I know routinely uses positive and negative controls to ensure the validity of results. That's standard practice even in labs that work exclusively with non-human DNA; it's particularly vital when human genetic material is involved.

It's a very simple process. You just add two samples to your experiment. One sample gets DNA from a known source - that's the positive control. A second sample gets - if you do everything right - no DNA at all. That's a negative control. You run the experiment. If you don't find DNA in the positive control, you did something wrong. If you do find DNA in the negative control, you did something wrong. If your positive control comes back positive, and your negative control comes back negative, there's at least a chance that you did everything right.

The lab that analyzed the DNA in these cases probably was doing controls, and doing it correctly, but they shouldn't have been the only ones to use controls. It's not hard for a forensic tech to use positive and negative controls in the field. If you want a positive control, you use the same collection materials that you use for the real samples, but grab something with your own DNA - one of your hairs, or a swab from your own cheek, whatever. For your negative control, just unwrap the collection swab and drop it right into a sample container. Treat those samples exactly as you would every other sample you collect. Give them sample numbers, send them to the lab, the works.

If a negative control comes back positive for DNA, try to match it anyway. If it matches you, you know you're the one who screwed up. If it matches things found at unrelated crime scenes, there's either a problem at the lab or the firm that makes the sample collection supplies. I find it impossible to believe that the German police could have been doing that, given how long it took them to figure out that they had a contamination problem.

On the bright side, though, it should be fairly easy for the manufacturer to figure out who they need to fire.

(Hat tips to revere and Jonathan Turley)

7 responses so far

  • Andrea M says:

    Oh come on, it is most CLEARLY not a problem with the controls. The woman obviously killed the first victims, then made sure to be hired at the cotton swab factory in order to fabricate her own alibi. We are talking about a clever fiend, here!

  • Bob O'H says:

    So what's the deal with Bavaria? I can work out most of it, but why were they using different swabs?
    Well, other than that they got their swabs from Swabia, presumably.

  • phisrow says:

    Somehow, I'm sure, this will become an argument for expanded powers...

  • Anonymous says:

    Negative controls are of limited use if less than one out of 500 samples is contaminated. The phantom's DNA was found in 40 cases in Germany, Austria, and France. The manufacturer now says they delivered about 20000 tubed cotton swabs to police. Sources differ about whether only to German or all police, and whether over the years or last year alone. In any case, 20000 is the minimum total number. And when suspicions on the swabs emerged, a batch of 300 further negative controls was done, without a positive result.
    You have to use different methods of quality management here, control is not sufficient. Much could be said about this, but buying swabs certified DNA-free instead of only sterile is the main thing from the point of view of the police. The rest then mainly is the manufacturer's problem. This they can certainly solve, but that costs money.
    In this case, the company supplied sterile swabs for 10 to 15 cents a piece, not knowing what they were to be used for. They were clearly not suitable for DNA analysis. An other german company which offers both types of swabs says their sterile ones cost 18 cents and DNA-free ones cost 15 times as much, which would be about 2,70 Euros a piece. This might be part of the explanation. And then, obviously sterile swabs worked quite well 15 years ago, because they are of course quite clean and polymerase chain reaction at that time was by far not as sensitive as it is today. "We've always been doing it this way, and it's always worked, hasn't it?"
    By the way, the phantom has been identified. It's a worker in the plastics manufacturing company where the tubes are produced and then assembled with the swab. This is done manually, using gloves and some protective clothing, though. Funny thing is this company is in Bavaria, but supplies the tubes to a medical equipment company in Baden-W├╝rttemberg which then radiation sterilizes the whole thing and which is itself a subsidiary of an Austrian company. The phantom's DNA was found mostly in Baden-W├╝rttemberg and parts of Austria, but never in Bavaria which is exactly in between.
    There you have it: Local procurement, business as usual, and no consideration on quality whatsoever.
    sources for the numbers (German):
    http://www.stuttgarter-zeitung.de/stz/page/detail.php/1984730
    http://www.spiegel.de/panorama/justiz/0,1518,615969,00.html

  • John says:

    There's a much wider problem here, not confined to the German police. Blind faith in the value of DNA evidence.
    There was ample evidence that there was no female serial killer. If two bits of evidence conflict, you either try and reconcile them, or assume that one is false. The second alternative was ignored in favour of the first. You have a photofit of a man with a beard? No problem, he's a transsexual.
    It's not just the police either. I first read of this two days ago, and have since read many reports of the case on the Internet. Everybody, journalists, and commenters uncritically accepted that the Phantom really existed. Despite ample reason to form the hypothesis of contamination.

  • compn says:

    its obvious the germans were onto carmen sandiego...

  • Roger says:

    I realise this thread has been dead for a year, but just in case you are interested: it has come out that something else that helped to delude the police was the criminal profiling.
    It is increasingly looking as though "profiling" is junk science [1]. But at any rate, in this case the profiler looked at all the totally contradictory modi operandi, and rather than saying "this makes no sense, this can't be just one person" instead constructed a bizarre composite that jammed in the features of the dozens of different cases.
    This apparent "professional" blessing encouraged the investigators to ignore their concerns about the obvious contradictions.
    ___
    1. Indeed, the Skeptics Association is beginning to run some debunking articles on it. To begin with, they have covered many of the famous "success stories" of profiling, which when looked at closely invariably turn out to be nothing but hindsight bias.