CSI lies and suspicious science

Or, Don't Try This in Your Home Lab, Kids!

Forensic science has come a long way since Sherlock Holmes bragged that he could identify 140 types of tobacco from their ash. And far be it from me to diss one of my favorite shows on the TV I don't own, the original CSI, which is loaded with fantastic sciency goodness, even if it is a little unrealistic. CSI? Unrealistic? Hate to break it to you kids, but, yeah. At the very least, the speed with which our intrepid heroes get their results would make any cop, ADA, or defense attorney double over in laughter, when they're not crying. DNA rape kits, appallingly, have as much as a six year backlog, according to a recent NY Times Op-Ed. That's not just because of the volume of cases involved, but the time it takes to analyze and match DNA samples.

 One of the reasons there's such a huge backlog is because of the complexity of testing and matching DNA. The molecule, as many of you may know, is a double helix composed of pairs of four "bases" (nucleotides): adenine, cytosine, guanine, thymine forming "ladder rungs" up the middle. Because of their chemical makeup, adenine only bonds with thymine and only cytosine bonds with guanine. Because there are millions of base pairs in every strand of DNA, the order in which these pairs bond on the molecule can be used to "fingerprint" both species and individual characteristics. But also because there are millions of base pairs, it's most useful (and quicker) to use known repeating pairs which are known to vary among individuals. Analyzing these can tell whether a DNA sample matches one from the same person, relations, or a complete stranger, not absolutely, but within a percentage of probability.

The actual fingerprinting is a four-step process, starting with the Southern Blot test and progressing through a radioactive probe, hybridization and the most accurate step, comparing Variable Number Tandem Repeats (VNTRs). The Southern Blot is where the DNA molecule is squeezed out of the cells, cut into pieces, and "unzipped" (denatured). Then it's nicked and marked with radioactive material to label the sequences. Each of these steps have their own individual processes, so a single analysis of one sample is a long process that can go wrong at any one of these steps. Then there's the kits used to gather the evidence. One of the biggest problems with DNA evidence is that it degrades over time, so the longer they sit in storage, the less chance there is of an accurate analysis. That's why that six-year backlog is worrisome. So that DNA matching thing? It's not perfect.

 Meanwhile, that cool favorite of our CSI heroes, Luminol, the chemical that makes blood long scrubbed away show up in dramatic blacklight dayglo, turns out to be susceptible to false positives from household substances as common as Drano crystals and bleach. A competing product called Hemaglow, while easier to use in the field and not as sensitive to false positives, is also less sensitive to diluted blood than Luminol. I actually first heard of the false positives problem in a mystery novel called In the Woods by Tana French (never underestimate what you can learn from fiction). But false positives are a problem in almost any chemical test. False positives turn up because the test is reacting to some substance that's similar to the one it's testing for, even if it's just the tag end of a completely different molecule. Field tests tend to be not very sensitive, for practical reasons, since accurate chemical analysis requires controlled conditions and technical knowledge. Most field tests are simple to use, but more like a dragnet than a fine-mesh seine.

For instance, the field test for the date-rape drug GHB turns up false positives to (wait for it . . .) soap. The test is simple to use: put the suspect substance in a pouch, seal it, break the enclosed ampule and watch the pretty colors. Check out this video, in which the drummer for the Germs, Paul Bolles, is busted for alleged possession of GHB, er, soap. (Once again, truth is stranger than fiction. Or at least more ironic.) You'll also get a quick lesson on how good soap is made.

While we're on the subject of field tests, what about those Breathalyzers? The good news (for drinkers) is that they're notoriously bad. The bad news (for law enforcement) is that . . . they're notoriously bad.  I'll let David J. Hanson, Ph.D., Professor Emeritus of Sociology of the State University of New York at Potsdam sum them up:

A major problem with some machines is that they not only identify the ethyl alcohol (or ethanol) found in alcohol beverages, but also other substances similar in molecular structure. Those machines identify any compound containing the methyl group structure. Over one hundred compounds can be found in the human breath at any one time and 70 to 80 percent of them contain methyl group structure and will be incorrectly detected as ethyl alcohol. Important is the fact that the more different ethyl group substances the machine detects, the higher will be the false BAC (blood alcohol concentration) estimate.

The real problem with Breathalyzers is that they only estimate your BAC; only a real blood test can provide a truly accurate reading, so demand one if you're pulled over. Because there are so many compounds that contain that methyl group structure, it's not that hard to get a false positive, not just from your own breath, but from environmental factors. It depends on where the test is administered and what local pollutants are in the air, and whether you've been, say, spray painting something in the last hour or so. It can also depend on your own body chemistry.

One of the compounds containing the methyl group structure is acetone (dimethyl ketone), not uncommonly found in the breath of diabetics and some low-carb dieters. In addition, the symptoms of hypoglycemia (and strokes or seizures) can mimic the symptoms of intoxication. It's not unheard of for people suffering hypoglycemia to be pulled over and accused of DWI. Driving while hypoglycemic is probably just as dangerous as DWI, but not criminal.

Which brings us to a newish and disturbing trend in law enforcement: suspicion of science. i09's Annalee Newitz posted a story about about a University of Saskatchewan chemistry major who was busted on suspicion of having a home meth or bomb lab in his parents' garage. The cops later decided it wasn't a meth lab, but that he might be making explosives with the chemicals they found. Like, oh, fertilizer (which, granted can be used in both meth production and explosives, but only in large quantities). Lots of other stuff commonly found in your home goes into meth production, and you certainly don't need lab ware to cook the stuff; a coffee maker will suffice. Which makes me glad I'm a tea drinker; I'd hate to be busted for owning a Mr. Coffee (and read the comments here, if you think I'm kidding). Or too many antihistamines.

Newitz rightly points out the similarity between this case and Steve Kurtz's, in which the FBI could not tell the difference between petri dishes of harmless bacterial cultures sitting in the open air and biological weapons. Initially facing 20 years in prison for making art that was critical of genetically modified food, Kurtz has finally had his stash of commonly available lab components and living art projects returned by the FBI and been cleared of all charges. Both of these cases point to a certain amount of scientific ignorance on the part of law enforcement officials, the same kind that sent the artist's books of Edith Kollath's into the black hole of the FBI's evidence room because they contained electronic devices and she was hand-carrying them onto a plane to return home to Germany. The electronic devices? Little engines inside the books that made them appear to breathe by pushing their covers and pages up and down. I suspect these were not complicated devices. How hard would it be to tell they weren't detonators?

So there's some cognitive dissonance going on not just in law enforcement but in our culture as well. Law enforcement has increasingly relied on science to catch and convict lawbreakers from drunk drivers to terrorists; at the same time, many law enforcement officials on the ground have a hard time drawing a line between criminals and scientific dabblers, whether they're artists, basement inventors, or the merely curious. It also has a tendency to over-rely on the accuracy of techniques that are meant to provide field yardsticks rather than defense-proof  evidence. A little more knowledge about mitigating factors and some general scientific knowledge wouldn't be amiss. We like what science does, but we don't really trust the people who do it.

UPDATE: I just ran across a study "conducted by the Federation of American Scientists (FAS) and the Federal Bureau of Investigation (FBI) revealed that while scientists are disposed to assist in criminal investigations, they often fear working with law enforcement agencies." One of the key findings:

Scientists feel that the FBI does not work well with the scientific community, specifically that law enforcement officers don't understand their work (76%), that these agencies are more interested in restricting research for security purposes than they are in the scientific value of the work (71%), that officers have an overzealous approach to security issues [emphasis mine] and an interest in censorship (63%), and that research will be restricted from publication (55%).

Surprise! Not, in light of the episodes I cited above.

But it's not all bad news from the world of forensics. There's a new fingerprint test that's showing some success in lifting even prints that have been wiped away, at least from things like shell casings. It works on the corrosion caused by sweat rather than by "developing" the residue of the secretions themselves, which can be wiped away. The salts left behind by sweat are heated with an electrical charge after being coated with a powder that reacts to the ionic salts that cause corrosion and are left behind by sweat, even after wiping. Unlike fingerprint secretions, corrosion is forever and happens on a micro leve whenever we touch metals that react to our body chemistry, like brass. Watch the National Geographic video about it. And remember to leave your gloves on when you pick up those shell casings. Oh, and rinse out that Mr. Coffee pot really well, too.

Comments

[Peggy]

I suppose the next thing you'll tell me is that the 3D hologram facial reconstructions on Bones aren't commonly used in forensics labs either.

More seriously, I think that the general ignorance about science coupled with the need for prosecutors and the Feds to look like they are actively doing something about criminals and terrorists (or "terrorists") has lead to some of the miscarriages of justice that you mention. I can understand being wary of bacterial cultures being grown in someone's kitchen, but I would think simple testing should be able to quickly determine whether confiscated bacteria or chemistry equipment or electronic gizmos are dangerous or not. But it seems like those tests are only done if law enforcement decides its in their best interest. And that's scary.

I also think it's outrageous that it's taking so long for the rape kits to be processed - crime labs clearly aren't being given the resources they need.

[Zack]

Some years ago I agreed to have my bathroom gone over by a team of police detectives; the previous tenant of the apartment was their prime suspect in a murder case, and the body had been dismembered, so they were looking for bloodstains.

The chemical test they used for this was very sensitive (as it needed to be, since the actual murder had happened some years ago) and it got false positives on any sort of iron oxide. So when they came to the rust stains on the ancient, crumbling sink, it lit up, and then the three of them all went into a little huddle -- "That looks like pipe rust to me. Do you agree?" "Yeah, that's rust." "Yup." and they marked it down on their chart as such.

It was very not-CSI.

[Igor Carron]

With regards to this pressure from law enforcement, here is another one for you:

http://www.boingboing.net/2008/12/29/chemistry-student-wr.html

or also here:

http://12angrymen.wordpress.com/2007/10/31/endangered-species-the-chemistry-set/

Beyond Chemistry, one of the issue I find fascinating is the inability of a large part of the population (that includes engineers) to also misunderstand what image processing can do. In many CSI episodes, image processing tasks such as tracking, deblurring or face matching look like a simple photoshop session. The hard work of calibration and so forth is not even hinted on.

Here is an analysis of Numb3rs : http://homepage.smc.edu/nestler_andrew/numb3rs.htm

Igor.

[Chris]

Freakonomics had a good post on DNA probabilities http://freakonomics.blogs.nytimes.com/2008/08/19/are-the-fbis-probabilities-about-dna-matches-crazy/

[x-ray fluorescence]

Hi,
The chemical test they used for this was very sensitive (as it needed to be, since the actual murder had happened some years ago) and it got false positives on any sort of iron oxide.

[AA]

There actually was an episode of CSI where they dealt with the fact that you can get false positives on things like bleach with luminol.