The Icepick Surgeon

by Sam Kean

  She got outstanding science grades at Boston Latin, and later claimed she’d graduated summa cum laude from there, even though the school didn’t grant such honors. She also told people, falsely, her parents were both doctors. These petty lies continued in college and then at the vaccine company and state drug lab, where she invented elaborate titles for herself, like “on-call supervisor for chemical and biological terrorism” and FBI “special agent of operations.”

  Still, however distasteful, the lies to this point hadn’t harmed anyone. But small lies have a way of accumulating momentum, and things soon took a dark turn.

  Dookhan’s lab identified drugs that the police seized during raids. Sometimes these were blocks of pure drugs; sometimes they were small quantities cut with baking powder or baby formula and divided into baggies or squares of tinfoil for sale on the street. Since many drugs look alike, the police would drop them off at the lab so Dookhan and her colleagues could identify them, which they did through a series of tests.

  The first round of tests, called presumptive tests, told the analysts the general class of drug they were dealing with. One test involved adding formaldehyde and sulfuric acid to an unknown powder. If the sample turned red-purple, it was an opiate; if it turned burnt orange, an amphetamine. Other chemicals might turn drugs green or blue.

  Let’s say the chemist has an opiate. She’d then run a second, confirmatory test to narrow her results down to a specific drug. The confirmatory test involved taking a bit of the unknown sample, dissolving it in liquid, and running it through some analysis inside a machine. Samples of known opiates (e.g., morphine, heroin, fentanyl) went through the same analysis on the same run. The machine then spit out several graphs—a sort of barcode for each sample. By comparing the barcode from the unknown sample to the barcode from the known samples, chemists could identify the exact drug involved and inform the police.

  Like drug labs nationwide, the Boston lab was drowning in samples to test. By 2003, their backlog had ballooned to several thousand items; the walk-in safe where they stored untested samples was eventually packed so full that it was considered a safety hazard to walk around inside. But with Dookhan’s arrival, things started looking up. She quickly distinguished herself as not only the hardest-working chemist (first to arrive, last to leave) but also the speediest. In her first year, she churned through 9,239 drug samples—three times the average of what the other nine chemists tested, and more than a quarter of the lab’s output overall. People there started calling her superwoman, a compliment that left her glowing. In emails to prosecutors that she worked with, she bragged about how indispensable she was to the lab.

  Privately, though, she was using this praise as a balm against pain. In 2004, she met an engineer from her native Trinidad and married him. Before long she was pregnant. But that first pregnancy ended in a miscarriage. (She later suffered another.) Each loss devastated her, and put a huge strain on her relationship.

  Drug analyst Annie Dookhan. (Courtesy of the Boston Herald.)

  Rather than take time off to cope, as her supervisor urged, Dookhan blotted out the pain by spending even more time at the lab bench. “I have chocolate and work,” she told her boss, “and that is my way of dealing with it.” The year after the first miscarriage, she set an even more torrid pace than before, racing through 11,232 samples, almost double the second-place chemist and four times the lab average. Dookhan did eventually give birth to a disabled son, which slowed her pace somewhat, but she continued to lap her fellow chemists year after year. Whereas most of them would check out two dozen samples to test at any one time, Dookhan usually took five or six dozen, and once took 119.

  Gradually, however, her coworkers grew suspicious about her superwoman pace. Some of this was common sense. How on earth could anyone work so fast? There were circumstantial clues as well. A colleague once caught Dookhan not calibrating her scale—a vital step to ensure accuracy, since the difference between, say, 27.99 grams and 28.00 grams of a drug meant a difference of several years in jail. Colleagues also noticed that, despite all the tests Dookhan recorded doing, she never actually seemed to use her microscope. In a related concern, she didn’t seem to generate enough trash. During one test, called a crystal test, chemists mixed an unknown drug with a liquid on a glass slide. Crystals would soon form. Different drugs made differently shaped crystals, which chemists identified under a microscope. Each test required a clean glass slide to avoid contamination, so based on the number of tests run, chemists should be throwing away a certain number of slides each month. Dookhan wasn’t. Colleagues peeked into her discard bin and noticed how bare it looked.

  Dookhan’s colleagues were right to be suspicious. Although it’s not clear when exactly it started, she was committing fraud on a massive scale. Instead of actually running tests, she “dry-labbed” her samples—simply glancing at them and guessing what they were.

  She got away with this by exploiting a flaw in her lab’s workflow. For chain-of-custody reasons, all drug samples were accompanied by “control cards,” records that indicated when the drugs were seized, what the police assumed the drugs were, and so on. This is good police procedure. The problem was, chemists like Dookhan had access to the control cards and could therefore see what drugs the police suspected. Allowing chemists to see this information was a bad idea anyway. Suggestions inevitably bias us, nudging us toward certain conclusions and away from others. Dookhan, however, outright exploited the flaw, using the police guess as her entire “analysis.” If they said it was heroin, it was heroin. No muss, no fuss.

  To be fair, Dookhan always tested unknown samples, those lacking control-card information, since she would have been guessing blindly. She also ran a full array of tests on roughly one-fifth of her samples, just to make sure. But otherwise she skipped all the bothersome chemistry and simply rubber-stamped things, to keep her numbers high. Equally bad, she’d then sign certificates claiming she’d run the tests and submit those to the police. These certificates served as evidence in courtroom trials, so she essentially perjured herself over and over.

  Now, in many cases, Dookhan’s dry-labbing made no practical difference: Cops generally know what drugs they’re seizing. So even though skipping the test violated a suspect’s right to due process, the final verdict probably would have been the same. But not always. And here’s where Dookhan strayed into truly sinful territory.

  Again, there were two rounds of testing at the lab. Often Dookhan would do the first round and another chemist the second, and sometimes the second round—the one that involved machines—would contradict Dookhan’s initial guess. In these cases, a retest was in order. But instead of claiming she’d made a mistake, which might put her superwoman reputation at risk, Dookhan would sneak off, find a pure sample of the drug she’d initially claimed, and submit that for retesting. Presto, the machine now gave the “correct” result. In other words, she started forging evidence to conceal her fraud.

  As a result, innocent people went to jail. One man was arrested with inositol, a white powder sold as a health supplement. Dookhan nailed him for cocaine. In another case, a drug addict tried to pull a rather foolhardy scam and sell a fragment of cashew to a fellow hophead, claiming it was crack. The hophead turned out to be an undercover cop. Still, it was just cashew, no big deal. The man then watched, stunned, as Dookhan swore in court to the contrary. “I knew she was lying” about running the tests, he later said. “Ain’t no way, no how, a cashew can turn into crack.”

  Not everyone Dookhan lied about went to jail; low-level drug offenders often didn’t. But drug convictions have consequences beyond prison terms. You can get deported or fired or kicked out of public housing. You can lose your driver’s license or the right to see your children. If you appear in court again, you’re also a repeat offender.

  Dookhan never gave a satisfying explanation for why she jeopardized so many people’s lives. Still, her words and actions do provide some hints. First, Dookhan seemed to enjoy busting drug dealers.
She was often inappropriately friendly with local prosecutors and wrote them earnest emails about getting bad guys “off the street.” One prosecutor offered to buy her drinks at a top-shelf bar. Another had to resign when his flirty emails with Dookhan became public. Once, she asked a prosecutor’s advice on whether she should even bother responding to a defense attorney’s plea for help with his client’s case.

  Dookhan was also severely stressed, which psychological research shows can tempt people to cut corners and act immorally. Given the huge backlog at the lab, everyone there faced substantial pressure to churn through samples. Compounding this problem, Dookhan had suffered multiple miscarriages and was unhappy at home; she had no family beyond her parents, and lived right next to her entire clan of in-laws, never an easy thing. That’s no excuse, but prolonged stress can deplete our mental stamina and lower our sense of empathy toward others. Given her own messy mental state, Dookhan might have found it easier to ignore the possibility that her fraud was ruining people’s lives.

  Especially when that fraud won her praise. Some people lie to manipulate others or gain material things. Dookhan wanted scientific glory—she loved being called superwoman. Her old supervisor at the vaccine lab also speculated that her status as an immigrant and a woman of color might have played a role. The supervisor, who is Black, said, “I understand what it is like to be a minority in America. I think that experience reinforced her determination to show that she was just as good, or even better.”

  Normally, that determination is a healthy thing, pushing people to achieve more and bust stereotypes. But Dookhan wasn’t earning her accolades; she pursued the glory without the underlying accomplishment. This is actually a common failing among those who commit scientific fraud. Rather than knowledge, they seek awards and prestige—the trappings of science rather than science itself. But it’s one thing to churn out fraudulent work in, say, optics or ornithology. Dookhan did so in a forensics lab, where people’s freedoms were at stake.

  Sadly, Dookhan was far from alone here. In recent decades dozens of other forensic scientists, at labs across the world, have been exposed as cheats and impostors. To critics, in fact, Dookhan’s case only reinforced the notion that forensic science itself was something of a fraud.

  In the United States, the roots of forensic science trace back to the Parkman murder case from chapter four, when doctors at Harvard Medical School used their anatomical expertise to nail John White Webster. Over the next several decades, forensic science expanded into arson investigation, firearm ballistics, and so-called impression analysis—the study of fingerprints, bite marks, footprints, splatters of blood, and the like. By the mid-twentieth century, forensic science was firmly established in the courts, and was viewed as a rational, objective alternative to the arbitrary and corrupt police work that reigned before.

  Unfortunately—and I hate to disappoint all you murder-mystery fans out there—much of forensic science is spotty at best and outright bunk at worst. In a damning report from 2009, the U.S. National Academy of Sciences outlined several glaring problems with forensic science—starting with the fact that most fields within it lack any scientific basis. Rather than being grounded in experiments and analysis, they’re merely a collection of hunches that have been gussied up in scientific jargon. As a result, different forensic experts often draw wildly different conclusions from the same sample. Hell, a single expert sometimes draws wildly different conclusions from the same sample at different times, depending on whether you mention beforehand that you think the suspect is guilty or innocent—strong evidence that bias drives the analysis.

  Equally damning is the lack of humility. Speaking from personal experience, something that drives science writers batty is the tendency of scientists to hedge everything; they’re always qualifying their statements and adding disclaimers about alternative explanations, even when the evidence seems strong. In contrast, many forensic experts—especially when testifying in court—boast of zero uncertainty. They claim they can match hair fibers or bite marks to someone with 100 percent accuracy, and do so 100 percent of the time. They project an aura of infallibility,1 and bluster their way through any questions that challenge their authority.

  To be clear, not all forensic science is garbage. Toxicology and pathology are solid, and the National Academy report singled out DNA analysis in particular as trustworthy. These fields have rigorous foundations and rely on well-grounded laboratory tests; and in the case of DNA analysis, it can reliably tie specific biological samples (e.g., blood or semen) to specific individuals. DNA analysts also routinely acknowledge uncertainty by attaching probabilities to their results. But most forensic fields do not meet these basic guidelines.

  Since the Academy report, the fields of fingerprint analysis and firearm ballistics have started to shore up their sloppy practices and shift toward scientific validity. And even the spottier forensic sciences could—if the practitioners would analyze evidence properly and show a little humility—find a valuable place in modern police work by adding weight to other testimony and supplementing an overall case. Until then, defendants will continue to suffer. By some estimates, “false or misleading forensic evidence” contributes to a quarter of all wrongful convictions in the United States, and some forensic disciplines have even poorer track records. In one study the FBI concluded that in 90 percent of cases involving microscopic hair samples there was “erroneous” testimony in court.

  Where does forensic drug analysis fit in? On the spectrum of validity, it’s closer to the DNA side of things. Drug tests are reliable and repeatable, and if they’re performed properly, they’re a solid part of a criminal case. If they’re performed properly.

  Dookhan’s downfall started with a whopper of a coincidence. In 2001, Boston police officers arrested a man named Luis Melendez-Diaz for dealing drugs outside a Kmart and hauled him off to jail in a police cruiser. On the way, the cops noticed Melendez-Diaz squirming in the back seat. Suspicious, they searched the car after booking him, and found several baggies of cocaine that he’d had on his body and had shoved into a partition in the back to discard them.

  The police happened to send the baggies to the very lab where Dookhan would soon find a job. By all accounts, the samples were processed properly, without any funny business. The chemist in charge signed three certificates stating that the drugs were cocaine, and this evidence helped convict Melendez-Diaz. All in all, a slam-dunk case.

  Except Melendez-Diaz’s lawyers put forth a novel argument. The Sixth Amendment to the Constitution states that “the accused shall enjoy the right… to be confronted with the witnesses against him” in court. Traditionally, this meant eyewitnesses, people who’d actually seen the crime committed. But Melendez-Diaz’s lawyers argued that forensic analysts should have to testify in person as well. In this case, because the lab chemist merely submitted certificates instead of appearing in court, the lawyers argued for throwing the conviction out.

  After appeals, the case wound its way to the Supreme Court in 2009. In a 5-4 ruling that defied the court’s usual partisan split (Ruth Bader Ginsburg joined Antonin Scalia and Clarence Thomas in the majority), the Supremes decided that Melendez-Diaz’s lawyers were correct: scientific analysts had to testify in court, to give the defendant a chance to challenge them. Partly this was a due process issue. The right to confront witnesses is essential to our notion of a fair trial, Scalia noted, and analysts therefore had to appear in court “even if they have the scientific acumen of Mme. Curie and the veracity of Mother Theresa.” But Scalia also suspected that not everyone in the drug labs was a Madame Curie or Mother Theresa. Some analysts were probably incompetent or even liars, he mused, in which case the “crucible of cross-examination” would expose them. He might as well have had Annie Dookhan in mind when he penned his decision.

  Now, there are good (and to my mind, convincing2) arguments against this ruling. But the upshot was that forensic drug analysts like Dookhan now had to appear in court to testify on a regular basis.


  So did cross-examination expose Dookhan, as Scalia predicted? Hardly. She kept right on lying on the stand. Dookhan would eventually testify 150 times in court, all under oath, and in all 150 cases she got away scot-free. The vaunted “crucible” of cross-examination failed to detect even the most egregious fraud in the history of forensic science.

  Still, the requirement to testify did help expose Dookhan in a roundabout way. Even though she and other analysts rarely spent more than twenty minutes on the stand, they often had to waste whole mornings or afternoons at the courthouse sitting around and waiting for their case to come up. Every hour at the courthouse was an hour that Dookhan couldn’t spend at her lab bench. As a result, her testing numbers plummeted. After the Melendez-Diaz ruling came down, she spent ninety-two hours testifying during the last six months of 2009, and managed to get through “only” 6,321 samples that year. The numbers for other analysts dropped as well, to an average of roughly 2,000.

  But here’s the thing. Over the following year, the other chemists’ numbers remained low. Dookhan’s didn’t. Call it bravado or sloppiness, but she spent 202 hours testifying in court in 2010, and nevertheless claimed to have churned through 10,933 samples— five times the lab average, and almost as high as her pre–Melendez-Diaz peak.

  This was the point at which her fellow chemists really got suspicious, and began tracking Dookhan’s time at the microscope and monitoring her discard bin. Around this time, Dookhan also got caught skipping important calibration checks on different machines, presumably to save time. Even worse, she got caught forging another coworker’s initials on some paperwork, in an effort to cover up the fact that she’d skipped steps. Indeed, some colleagues later wondered if she’d actively wanted to get caught, given how flagrant her violations were.