Tens of thousands of marijuana-related convictions have rested on the results of a widely used test that is infamous among scientists for its unreliability
By John Kelly
RAISED in Montana and a resident of Alaska for 18 years, Robin Rae Brown, 48, always made time to explore in the wilderness. On March 20, 2009, she parked her pickup truck outside Weston, Florida, and hiked off the beaten path along a remote canal and into the woods to watch birds and commune with nature. "I saw a bobcat and an osprey," she recalls. "I stopped once in a nice spot beneath a tree, sat down and gave prayers of thanksgiving to God." For that purpose, Brown had packed a clay bowl and a smudge stick of sage, sweet grass and lavender that she had bought at an airport gift shop. Under the tree, she lit the end of the smudge stick and nestled it inside the bowl. She waved the smoke up toward her heart and over her head and prayed.
As darkness approached, she returned to her pickup truck to find Broward County's Deputy Sheriff Dominic Raimondi and Florida Fish and Wildlife's Lieutenant David Bingham looking inside the cab. The two men asked what she was doing and when she said she had been bird watching, Bingham asked whether she had binoculars. As she opened her knapsack, Raimondi spotted her incense and asked if he could see it. He took the bowl and incense, asking whether it was marijuana. "No," she recalls saying. "It's my smudge, which is a blend of sage, sweet grass and lavender." "Smells like marijuana to me," said Raimondi, who admitted he had never heard of a smudge stick. He then ordered Brown to stand by her truck while he took the incense back to his car and conducted a common field test known as a Duquenois-Levine, or D-L, test. The result was positive for marijuana.
Brown protested, telling them the smudge was available for purchase online for about $7 and giving them the name of a website that sold it—information Bingham used his laptop to verify. But the men still searched her truck. After an hour and a half they finally allowed Brown to go home and told her that if a lab test confirmed the field test results, a warrant would be issued for her arrest.
Exactly 90 days later, Brown was arrested at the spa in Weston, Florida, where she has worked as a massage therapist for three years. She was handcuffed in front of clients and co-workers and charged with felony possession of marijuana. She was brought to a local police precinct in the town of Davie where she was booked and held for three hours. Unable to post the $1,000 bail, she was transferred to the Women's Correctional Facility in Pompano Beach. At no time was she read her rights.
Five hours after her arrest, she was finally allowed a brief phone call and left a message for her boyfriend to post her bail. At the jail, a female officer came in and told Brown to take off all her clothes. She had already been searched at the precinct station and had her shoes, socks and bra confiscated. "I'm on my period," Brown said. "I don't care," said the officer, who ordered her to pull her underwear down to her ankles, squat over the floor drain and cough. The following morning at 4:30am Brown was released onto the streets of Pompano Beach with no idea where she was.
The next day, Brown found a lab and submitted to voluntary hair and urine tests. These came back clean. She had previously worked for 16 years as a transportation systems specialist with the Federal Aviation Administration, a job that required airport security clearances, so drug tests were nothing new to her. During those years, she was frequently required to pass random drug and alcohol tests.
She later learned that her incense had never been subjected to a confirmatory lab test. She had been arrested and jailed solely on the basis of her positive D-L test results.
Reefer Test Madness
The Duquenois test was developed in the late 1930s by a French pharmacist, Pierre Duquénois, while he was working for the United Nations division of narcotics. In 1950, he completed a study for the U.N. that claimed that his test was "very specific" for marijuana; it was adopted by the U.N. and crime labs around the world as the preferred test for marijuana.
After undergoing several modifications, including the use of chloroform, the test became known as the Duquenois-Levine test, and became widely popular. Though scientists would show in the 1960s and 1970s that the D-L test was nonspecific, meaning it rendered false positives, it remains today the most commonly used test for marijuana—employed in many of the 800,000 marijuana arrests that take place each year.
The test is a simple chemical color reagent test, easy to perform but difficult to interpret. To administer the test, a police officer simply has to break a seal on a tiny micropipette of chemicals and insert a particle of the suspected substance; if the chemicals turn purple, this indicates the possibility of marijuana. But the color variations can be subtle, and readings can vary by examiner.
The field test kits are produced by a variety of manufacturers, the most popular brands being NIK and ODV. Literature about the D-L from NIK's makers states that it is only a "screening" test that "may or may not yield a valid result" and may produce "false positive results." Yet since at least 1990, arresting officers, with the support of prosecutors, have regularly bypassed lab analysts and have purported to identify marijuana at hearings and trials only on the basis of visual inspection and the nonspecific D-L field test. And the manufacturers have taken note.
In 1998, ODV reported in its newsletter that a growing number of police departments were using its D-L field test, marketed as the NarcoPouch, as "their sole method of testing and identifying Marihuana [sic] ... To have Officers properly trained in identifying Marijuana and taking the Crime Lab out of the loop is a tremendous cost saving venture for the State ... and gives the individual Officers testing the material a greater sense of satisfaction in completing their own cases." NIK, too, argued that depending exclusively on D-L field tests saves time and money. "Crime laboratories are so busy that drug tests take too long," NIK states on its website. "With the cooperation of the Prosecuting Attorney, many police agencies have turned to presumptive drug testing. If the results indicate that an illegal substance is present, criminal charges may be filed."
Experts in Four Days
In June 2006, the Virginia legislature went so far as to pass "emergency regulations" permitting law enforcement officers to testify at trial for simple possession of marijuana cases solely on the basis of a D-L field test. Prior to these regulations, officers had to send suspected material to an approved lab for testing. Nothing in the new legislation specified that the field tests used had to be specific, or even accurate. Frederic Whitehurst, a North Carolina–based defense attorney and former FBI special agent with a doctorate in chemistry, considers the law to be an unconstitutional usurpation of the authority of the courts to determine what test results can be admitted as valid evidence.
The trend toward police officers using the D-L as a confirmatory test has been encouraged by the National Institute of Justice, an agency of the Department of Justice that has funded programs to transform police officers into court experts based on their use of these faulty field tests. One such ongoing program for the Utah police claims to offer, in four days, "the necessary training" to positively identify marijuana, which would allow officers to serve as "expert witnesses in the courtroom setting." The program briefly covers the "botany, chemistry and analysis of marijuana preparations," after which police officers, including street detectives and crime-scene lab personnel, "will assume responsibility for all of their agency's marijuana submissions." By the end of 2005, such submissions became the exclusive provenance of the Utah officers who had attended the training, and suspected marijuana samples were no longer accepted at the state lab for processing.
In 2009, the Georgia Bureau of Investigation trained more than 1,600 police officers in the use of the D-L test, resulting in a 98 percent reduction in the use of marijuana lab tests. This troubling program garnered the bureau a 2009 Vollmer Excellence in Forensic Science Award by the International Association of Chiefs of Police.
Test 'Should Never Be Relied Upon'
Despite its widespread use, as early as the 1960s, the D-L test had been proven incapable of definitively identifying the presence of marijuana in a seized substance. A 1968 article in the Chemical and Pharmaceutical Bulletin of Japan reported that the D-L tests "lack in adequate specificity." In 1969, M. J. de Faubert Maunder, a chemist in the Ministry of Technology, a British government agency, reported finding 25 plant substances that would produce a D-L test result barely distinguishable from that of cannabis and cautioned that the D-L test "should never be relied upon as the only positive evidence."
Several articles in the Journal of Forensic Sciences further disproved any claims that the test could specifically identify marijuana. A 1972 study found that the D-L test would test positive for many commonly occurring plant substances known as resorcinols, which are found in over-the-counter medicines. For instance, Sucrets lozenges tested positive for marijuana. This study concluded that the D-L test is useful only as a "screen" test and was not sufficiently selective to be relied upon for "identification."
In 1975, Dr. Marc Kurzman at the University of Minnesota, in collaboration with 14 other scientists, published a study in The Journal of Criminal Defense that concluded: "The microscopic and chemical screening tests presently used in marijuana analysis are not specific even in combination for 'marijuana' defined in any way." In the 35 years since that study was published, no one has ever refuted this finding.
Indeed, recent research has confirmed Kurzman's findings. In 2008, Whitehurst, the chemist and former FBI agent, substantiated Kurzman's findings in an article in the Texas Tech Law Review. That same year, Dr. Omar Bagasra, director of the South Carolina Center for Biotechnology, conducted experiments in his lab also demonstrating that the D-L test is nonspecific and renders false positives. Bagasra, too, has impeccable credentials—he's a leading pathologist and a board-certified forensic examiner.
A number of high courts have been persuaded by this evidence, and have found that the D-L test does not prove the presence of marijuana in a seized substance. In 1979, a trial judge in North Carolina blocked the marijuana conviction of Richard Tate, which was to be based on positive D-L test results. In this case, too, the trial judge found that the D-L test was "not specific for marijuana" and had "no scientific acceptance as a reliable and accurate means of identifying the controlled substance marijuana." On that basis, the judge allowed the defendant to suppress the use of the test results as evidence. This finding was upheld by the North Carolina Supreme Court.
Also in 1979, the U.S. Supreme Court in Jackson v. Virginia ruled that the results of nonspecific tests could not be the basis for prosecution or conviction. In other words, if the only evidence is a positive D-L test, then the case must be dismissed.
Big Hairy Problem
Shoddy science, though, has muddied the waters. Several studies claim, falsely, to have validated the specificity of the D-L test. The best-known D-L "validation" study, and thus the most damaging to defendants, was published in 1972 by John Thornton and George Nakamura in the Journal of Forensic Science Society. It instantly made the D-L test the gold standard across the country for marijuana identification. But the report is internally contradictory and scientifically flawed. On the opening page of the article, the authors state that the D-L test is a "confirmation" test for marijuana. Such a test must be capable of proving the presence of the drug beyond a reasonable doubt, specifically identifying the drug to the exclusion of all other possible substances and producing neither false positives nor false negatives.
However, the researchers' own findings contradict their conclusion and show instead that the D-L test merely screens for marijuana. The authors themselves reported that the D-L test gave false positives and was not a confirmatory test even when cystolithic hairs—visible on the leaves of marijuana and other plants—are found on the suspected substance. They claimed that "the Duquenois-Levine test is found to be useful in the confirmation of marijuana" when cystolithic hairs are observed "since none of the 82 species possessing hairs similar to those found on marijuana yield a positive test." The problem is, as the authors noted, there are hundreds of plants with cystolithic hairs that they did not test, making their sample of 82 species woefully inadequate. In effect, they admitted that the botanical exam itself was nonspecific. Combining two nonspecific tests does not make a specific, confirmatory test, as the D-L and the botanical exam both could easily render false positives.
Without having proved specificity, the authors nevertheless claimed it. They also noted its widespread use as if it were proof of its efficacy, mentioning that the D-L test was adopted as a preferential test by the League of Nations Sub-Committee of Cannabis and that a version of the test was proposed by the United Nations Committee on Narcotics as a specific test for marijuana. (The U.N. subsequently found that only gas chromatography/mass spectrometry analysis could affirmatively identify marijuana.)
Inexplicably, this Thornton-Nakamura study is cited by the Drug Enforcement Administration and labs around the country as justifying the use of the D-L test alone or in combination with the microscopic visual exam for proving the presence of marijuana in a seized substance. Even some courts have erroneously ruled that the D-L test is specific and confirmatory. The most egregious example occurred in 2006. U.S. District Judge William Alsup found the D-L test to be a specific identification test and declared, grandiosely: "Despite the many hundreds of thousands of drug convictions in the criminal justice system in America, there has not been a single documented false-positive identification of marijuana or cocaine when the methods used by the SFPD [San Francisco Police Department] Crime Lab are applied by trained, competent analysts." In fact, according to an affidavit in that case from a senior criminologist at the SFPD, its lab had, for 40 years, used the D-L test in combination with a botanical exam to identify marijuana—two nonspecific tests that can each produce false positives. (A spokeswoman says that current SFPD policy is to subsequently confirm these results with gas chromatography/mass spectrometry.)
Challenging the Test
Robin Rae Brown never even faced trial on marijuana possession charges. After she was released from jail, she retained this author as a defense expert. When I first spoke with her attorney, Bill Ullman, he had never heard of the D-L test and said he normally plea-bargained cases like Brown's. I urged him to challenge the test and provided him with several scientific studies cited in this article, relevant court decisions, including Jackson v. Virginia and other information. When Ullman made inquiries, he discovered that the sheriff's department had never performed a lab test to confirm his field test results. Brown, he discovered, had been charged with a felony solely on the basis of the D-L test and Officer Raimondi's "opinion."
At Ullman's insistence, the sheriff's department finally performed a gas chromatography/mass spectrometry (GC/MS) analysis on Robin's smudge, which came out negative. State Attorney Berki Alvarez immediately dropped the charges against her, noting to Ullman, "the scariness that a person could be arrested under such conditions."
Even scarier was the lab's revelation that it does not conduct GC/MS analysis until just before a trial, as most marijuana possession defendants plea bargain before the trial. If Brown had accepted a plea bargain, she would have been wrongfully convicted and saddled with a criminal record that could have damaged her future job prospects. How many others before and since have accepted plea bargains based on false positives from a D-L test?
"I am just now willing to share this story," Brown wrote months after her arrest, "because it was embarrassing and I didn't want to worry my family and friends." After some serious thought, she recently decided to file a lawsuit for wrongful arrest. "I would like to see them stop using the bogus field tests and to improve their procedures at the county crime lab," she says. "I would like the public to be aware of the faulty field tests."
In truth, everyone arrested on marijuana charges has a Constitutional right to a GC/MS analysis. Otherwise, they are being denied both due process and a fair trial. "It is not only unnecessary for the courts to continue to accept conclusory drug identifications based on nonspecific tests, it is also unwise for them to do so," wrote Edward Imwinkelried, a professor of law at the University of California at Davis whose work on scientific evidence has been cited by the Supreme Court. Sustaining evidence from nonspecific tests like the D-L, he concludes, "is both bad science and bad law."
John Kelly is a court-certified expert witness on drug tests and author of 'False Positives Equal False Justice' and the forthcoming book, 'How to Obtain a Pretrial Dismissal of Marijuana Charges or an Acquittal.' He can be contacted at: firstname.lastname@example.org. Research support for this article was provided by The Investigative Fund of The Nation Institute.
Send letters to the editor here.