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 bird watch 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, Robin had packed a clay bowl and a “smudge stick,” a stalk-like bundle of sage, sweet grass, and lavender that she had bought at an airport gift shop in Albuquerque, New Mexico. 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. Spiritual people from many cultures, including Native Americans, consider smoke to be sacred, she told me, and believe that it can carry their prayers to the heavens.

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, Officer 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 Robin 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.

Robin protested, telling them the smudge was available for purchase online for about $7 and gave them the name of a Web site that sold it — information Officer Bingham used his laptop to verify. But the men still searched her truck. After an hour and a half they finally allowed Robin 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, Robin 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 because she was not allowed to call her boyfriend Michael, 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 Michael to post her bail. At the jail, a female officer came in and told Robin 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,” she 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 4:30 a.m. she was released onto the streets of Pompano Beach with no idea where she was.

The next day, Robin 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.

The Preferred Test for Marijuana

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 UN which claimed that his test was “very specific” for marijuana; it was adopted by the UN 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 — used 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 with seeming satisfaction 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” (emphasis added). 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.

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 which 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 Chemistry and Pharmacy 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 UK government agency, documented the unreliability of the D-L test in an article in the Bulletin on Narcotics, noting that test results depended heavily on the subjective judgment of the analyst — and thus could easily vary dramatically from lab to lab. “[A] positive test is not recorded until this colour has been identified,” he wrote, “and because it is almost impossible to describe in absolute terms it is best recognised by experience.” Moreover, he reported finding twenty-five 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 1969 study in the journal reported false positive results from “a variety of vegetable extracts.” 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.” Still another study, in 1974, showed that 12 of 40 plant oils and extracts studied gave positive D-L test results.

In 1975, Dr. Marc Kurzman at the University of Minnesota, in collaboration with fourteen 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 1973, the Supreme Court of Wisconsin ruled that the D-L test “standing alone is not sufficient to meet the burden of proving the identity of the substance beyond a reasonable doubt.” The court specifically noted that the D-L field tests used in this marijuana possession case “are not exclusive or specific for marijuana.”

Similarly, 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, which found that D-L test “was not scientifically acceptable because it was not specific for marijuana” and thus “the test results were properly suppressed.”

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.

As noted, even the test’s manufacturers do not claim that their product can definitively identify marijuana. The literature accompanying NIK’s NarcoPouch 908 cautions, “The results of a single test may or may not yield a valid result… There is no existing chemical reagent system, adaptable to field use, that will completely eliminate the occurrence of an occasional invalid test results [sic]. A complete forensic laboratory would be required to qualitatively identify an unknown suspect substance.”

Shoddy Science

Shoddy science, though, has muddied the waters. Several studies claim, falsely, to have validated the specificity of the D-L test. For instance, a seemingly authoritative 2000 study funded by the National Institute of Standards and Technology (NIST) purported to have validated the capacity of the D-L test to specifically and definitively identify marijuana. The title of the article, published in Forensic Science International, “Validation of Twelve Chemical Spot Tests for the Detection of Drugs of Abuse,” misstated the researchers’ actual findings. In fact, the study’s authors found that the twelve tests it analyzed, including the D-L, were nonspecific. “The tests,” they wrote, “are not always specific for a single drug or class.” Speaking of the D-L test, they wrote that “mace, nutmeg and tea reacted with the modified Duquenois-Levine,” meaning that they produced false positives. They also noted, echoing Maunder’s 1969 article, that the D-L test is subjective: “The actual color…may vary depending on many factors [including] the color discrimination of the analyst.”

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 Journal of Forensic Science Society. It instantly made the D-L test the gold standard across the country for marijuana identification. But just like the NIST study, this report is internally contradictory and scientifically flawed. On the opening page of this 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 eighty-two 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: “The specificity of the Duquenois reaction has been established, empirically at least, over the past three decades. No plant material other than marijuana has been found to give an identical reaction.” 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 UN 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 forty 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.)

In March 2009, a committee of the National Academy of Sciences, speaking of the D-L and other tests, called the analysis of controlled substances “a mature forensic science discipline”; “one of the areas with a strong scientific underpinning”; and an area in which “there exists an adequate understanding of the uncertainties and potential errors.” These incorrect assertions relied on assurances from government witnesses that “experienced forensic chemists and good forensic laboratories understand which tests (or combinations of tests) provide adequate reliability.” The committee’s main witness was Joseph Bono, the former director of a regional DEA lab, who had previously issued a sworn affidavit, referring to the D-L and other forensic tests, which asserted that “tests and instruments that are properly used by qualified forensic chemists are incapable of producing a false positive.” But experience and competence cannot make a test specific if it is not — nor can they make it immune from false positives.

In 2008, Senator Jim Webb, D-VA, said, in announcing a proposed bill, that “the criminal justice system as we understand it today is broken, unfair.” This unfairness is visible every day in the disparate and contradictory court decisions regarding the admissibility of D-L test results. Not only have courts contradicted one another on admissibility, but some courts have even chosen to admit the results of a D-L test while ruling that it does not prove the presence of marijuana beyond a reasonable doubt. This patchwork of admissibility means that a person in one state can be convicted of possessing marijuana on the sole basis of the D-L test while a resident of another state cannot.

In 1978, the Supreme Court of Illinois in The People of the State of Illinois v. Peppe Park illustrated this confused, unconstitutional state of affairs. In denying the admission of ipse dixit (“It’s marijuana because I say it’s marijuana”) reports, the court found that “police officers may not be presumed to possess the requisite expertise to identify a narcotic substance…because it simply is far too likely that a nonexpert would err in his conclusion on this matter, and taint the entire fact-finding process.” This court cited a study that found 241 incorrect identifications of marijuana by arresting police officers. Yet in the same decision, the court erroneously claimed that “to determine accurately that a particular substance contains cannabis, all that is necessary is microscopic examination combined with the Duquenois-Levine test.”

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 Robin’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. Robin, 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 Robin 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,” Robin 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. “Conclusory drug identification testimony is antithetical and offensive to the scientific tradition, and courts should not allow ipse dixit to masquerade as scientific testimony… Even more importantly, sustaining such drug identifications places a judicial imprimatur on testimony that cannot justifiably be labeled scientific. The rejection of such identifications is necessitated not only by due process but also by the simple demands of intellectual honesty.”

Sustaining evidence from nonspecific tests like the D-L, he concludes, “is both bad science and bad law.”

This article was reported in collaboration with The Investigative Fund at The Nation Institute, now known as Type Investigations.