This Dog Can Send You to Jail

How cops and their canines manufacture probable cause


Robert J. Burns, a 55-year-old retired nurse who lives in St. Louis, was returning from a trip to the West Coast last October when his white Nissan pickup truck was pulled over on Interstate 40 near Amarillo. Burns was carrying a 12-foot aluminum fishing boat on top of the truck, and he had been struggling against high winds that kept pushing him toward the shoulder. The sheriff's deputy who stopped him thought he might be drunk.

"He asked me to step out and come back to his car," Burns says, "and that's when I noticed the dog in the back seat, a yellowish Lab. I explained that I hadn't been drinking and my getting on the shoulder of the road was strictly from the wind. He said that he was going to write me a warning, and I said, 'OK, that's fine.' He asked me if I had any drugs in the car. I said, 'No, sir, I don't do drugs, and I don't associate with people who do.' He asked me would I mind if he searched my vehicle, and I said, 'Well, yes, I would mind if you searched my vehicle.'"

But thanks to the U.S. Supreme Court, the deputy did not have to take no for an answer. In the 2005 case Illinois v. Caballes, the Court declared that "the use of a well-trained narcotics-detection dog…during a lawful traffic stop generally does not implicate legitimate privacy interests." So the deputy was free to walk his dog around Burns' truck. "He got out with this dog and went around the car, two or three times," Burns says. "He came back and said the dog had 'passively alerted' on my vehicle." Burns, who is familiar with drug-detecting dogs from his work as an M.P. at Edwards Air Force Base in the 1970s, was puzzled. Properly trained police dogs are supposed to indicate the presence of drugs with a clear, objectively verifiable signal, such as sitting down in front of an odor's source or scratching at it. Yet "the dog never sat down, the dog never scratched, the dog never did anything that would indicate to me that it thought there was something in there."

The deputy and another officer who arrived during the stop nevertheless went through Burns' truck for half an hour or so, reaching up into the boat, perusing his cargo, looking under the seats and the hood, examining the gas tank and the undercarriage. They found no trace of drugs, although they did come across the loaded pistol that Burns mentioned to them once it was clear they planned to search the truck.

"They were cool with the gun," Burns says. "If it had been California, God knows what would have happened." He was so relieved that he barely minded the delay and inconvenience, which stretched a brief traffic stop into more than an hour. "I'm not a lawyer, and I'm not a super-libertarian," Burns says. "Once I realized that the pistol was not going to be an issue, man, they could have spent all day going over that car and under that car. My only concern was that one of the guys might have slipped something in to cover up for the fact that they didn't find anything."

That's one way of looking at it. But even if you are neither a lawyer nor a super-libertarian, you might wonder 1) how often this sort of thing happens, 2) how it came to be that police can get permission from a dog to rifle an innocent man's belongings, and 3) whether that state of affairs is consistent with the Fourth Amendment's guarantee against "unreasonable searches and seizures." The answers, in brief, are 1) fruitless searches based on dog alerts happen a lot more often than commonly believed, 2) dogs acquired this authority with the blessing of credulous courts mesmerized by their superhuman olfactory talents, and 3) this dog license is hard to square with the Fourth Amendment, unless it is reasonable to trust every officer's unsubstantiated claim about how an animal of undetermined reliability reacted to a person, a suitcase, a car, or a house.

All of these issues come together in two cases the U.S. Supreme Court heard a few weeks after Bob Burns was pulled over. Florida v. Harris raises the question of how a judge knows that a dog's alert is reliable enough to justify a search. Florida v. Jardines asks whether police need a warrant to use a drug-sniffing dog at the doorstep of a home. These cases, which will be decided by this summer, give the Supreme Court an opportunity to reconsider its heretofore unshaken faith in dogs, or at least limit the damage caused by the amazing canine ability to transform hunches into probable cause.

'A Creature of Legal Fiction'

The foundational text of the courts' canine cult is U.S. v. Place, a 1983 decision involving an airport search that found a kilogram of cocaine in a suitcase to which a dog had alerted. The Supreme Court unanimously concluded that the Drug Enforcement Administration (DEA) violated the Fourth Amendment by keeping the bag for 90 minutes before presenting it to a dog. But instead of stopping there, Justice Sandra Day O'Connor, in a majority opinion joined by five of her colleagues, gratuitously ventured into an issue that had not been addressed by the parties to the case and did not need to be resolved for the Court to decide whether the seizure and search were legal. O'Connor opined that "a 'canine sniff' by a well-trained narcotics detection dog…discloses only the presence or absence of narcotics" and "does not expose noncontraband items that otherwise would remain hidden from public view." Because of this specificity, O'Connor concluded, "exposure of respondent's luggage, which was located in a public place, to a trained canine…did not constitute a 'search' within the meaning of the Fourth Amendment."

Two decades later, when the Court extended this principle to cars in Caballes, dissenting Justice David Souter noted that O'Connor's conclusion "rests not only upon the limited nature of the intrusion, but on a further premise that experience has shown to be untenable, the assumption that trained sniffing dogs do not err." In reality, Souter said, "the infallible dog…is a creature of legal fiction." Souter cited examples of dogs accepted as reliable by courts that had error rates of up to 38 percent. He added that "dogs in artificial testing situations return false positives anywhere from 12.5 to 60 percent of the time."

If anything, Souter gave drug-sniffing dogs too much credit. A 2011 Chicago Tribune analysis of data from suburban police departments found that vehicle searches justified by a dog's alert failed to turn up drugs or drug paraphernalia 56 percent of the time. In 1979 six police dogs at two public schools in Highland, Indiana, alerted to 50 students, only 17 of whom possessed contraband (marijuana, drug paraphernalia, and cans of beer), meaning the false positive rate was 66 percent. Looking at the performance of an Illinois state police K-9 team during an 11-month period in 2007 and 2008, Huffington Post reporter Radley Balko found that the dog sniffed 252 vehicles and alerted 136 times, but 74 percent of the searches triggered by those alerts did not find measurable amounts of illegal drugs. Similarly, a 2006 study by the New South Wales Ombudsman in Australia, an independent agency analogous to the U.S. Government Accountability Office, looked at more than 10,000 searches of people triggered by dog alerts and discovered that 74 percent of them found no illegal drugs. More-recent data from New South Wales indicate an even higher error rate: 80 percent in 2011.

Those numbers look almost respectable compared to the results of a 1984 operation in which Florida state police stopped about 1,330 vehicles at roadblocks and walked dogs around them. If one dog alerted, another was brought in, and vehicles were searched only if both dogs indicated the presence of illegal drugs. That happened 28 times, but those searches yielded just one drug arrest. In other words, even when two dogs both signaled the presence of drugs, they were wrong 96 percent of the time.

What is going on when dogs alert and no drugs are found? Police and prosecutors usually claim these are not really false alarms because the dog must have detected otherwise imperceptible drug traces left on clothing, cars, or personal possessions. "It's a convenient excuse," says Lawrence Myers, a veterinarian and neurophysiologist at Auburn University who is an expert on dogs' olfactory capabilities. While dogs can indeed smell traces of drugs that are no longer visibly present, he says, "no one knows how big that reality is." When police use drug residue as an all-purpose explanation for what appear to be erroneous alerts, Myers says, "the first term that comes to mind involves a male bovine and the ingestion of grass."

Consider how Christopher Jbara, a U.S. Border Patrol agent, explained an unsuccessful dog-triggered search observed by a Tucson Citizen reporter in 2008. "He said the car had most likely been contaminated on one side of the border or the other and it was likely the driver was not aware," the Citizen reported. "He said the car's windshield had been washed by a window washer on the street before crossing the border, and the water used to clean it could have been contaminated with bong water."

New South Wales Police Inspector Chris Condon tells a somewhat more plausible story. In response to the 2011 numbers indicating that his department's dogs were wrong four times as often as they were right, he told The Sydney Morning Herald that "80 per cent of indications by the dogs result in either drugs being located or the person admitting recent contact with illegal drugs." The implication is that in most cases where people were searched and had no drugs, they had recently smoked marijuana (by far the most common drug found in successful searches) or been around pot smokers, which is why they smelled suspicious to the police dogs.

But that supposition is impossible to confirm, and it is not even clear what Condon means by "recent contact." More to the point, the likelihood of actually finding evidence of a crime is the relevant consideration (in Australia as well as the United States) in determining when police may search someone, meaning a dog's alert can justify a search only if it indicates that drugs are currently present.

'They Can Say Whatever They Want to Say'

The issue of what counts as a false alarm is central to Florida v. Harris. The defendant, Clayton Harris, was pulled over twice in 2006 by Officer William Wheetley of the Liberty County Sheriff's Office, once for an expired tag and once for a malfunctioning brake light. On both occasions, after Harris declined to let Wheetley search his pickup truck, the officer walked a German shepherd named Aldo around the vehicle. On both occasions, Wheetley reported, Aldo alerted by getting excited and sitting down in front of the driver's side door handle. And on both occasions, Wheetley searched the truck without finding any substance that Aldo was trained to detect. But during one of the stops, Wheetley found 200 pseudoephedrine tablets, along with other chemicals and supplies used to make methamphetamine, which led to Harris' arrest. Harris pleaded guilty to possession of a listed chemical with the intent to unlawfully manufacture a controlled substance, a second-degree felony punishable by up to 15 years in prison, but reserved the right to challenge the legality of Wheetley's search.

Since Wheetley did not find any illegal drugs in the truck and Aldo is not trained to detect pseudoephedrine, what are we to make of the alert Wheetley reported? He speculated that Aldo reacted to traces of meth left by Harris' hand, which might be true. Then again, Aldo might have smelled drug residue left by someone else—perhaps, as Harris' lawyer suggested, an addict looking for unlocked vehicles with stuff to steal. The dog's alert does not tell us who left the odor, or even which drug it was. Police dogs generally are trained to detect several substances, and they alert the same way to all of them.

Russ Jones, who worked as a K-9 officer and narcotics detective in San Jose, California, for 10 years and is now a member of Law Enforcement Against Prohibition, notes that the drug-residue excuse is a double-edged sword for police, because it undermines the case for using dog alerts to justify searches. "You're telling me that my car can be searched because the guy who changed the tires at the tire shop smokes marijuana, and his hands tightened up the lug nuts and put the hub cap back on?" Jones says. "Suppose the UPS guy uses amphetamine or cocaine, and he drops off a book that I ordered from Amazon.com. If a dog smells it, that gives you the right to search my home?"

Traces of drugs on the outside of Harris' truck are not the only possible explanation for the alert Wheetley reported. Perhaps Wheetley, who said Harris seemed nervous and restless, was so sure the guy was up to no good that he misinterpreted Aldo's behavior. If a dog handler "wants to see the alert," says Jones, "he sees it."

Alternatively, Jones says, "because [the handler] feels the guy is guilty, he just says the dog alerted and uses that as a pretext to search." Myers recalls a case on which he worked as a defense consultant where an officer claimed a dog alerted as he walked it around a suspect's car. In the dash-cam video of the stop, the dog was not visible, but the officer was. "When I had him questioned about how long it took the dog to alert, he said a few seconds," Myers says. "So there should have been at least a two-second pause in front of the car. Nope. There was no pause."

Jeff Weiner, a prominent Florida defense attorney who frequently deals with drug-sniffing dogs, says he commonly sees videos in which "someone will stand in front so you can't see the dog, and then you'll hear them say, 'Oh, the dog just alerted.' And then they'll step away." Weiner adds that many police departments have stopped recording K-9 teams at work "because they realized that the dogs don't alert when the cops say they alert." Without video, he says, "they can say whatever they want to say, and there's no way to challenge it."

Assuming Aldo really did alert to Harris' truck, he might have been reacting to Wheetley's suspicions. "If someone is acting quite twitchy and nervous," says Myers, "that evokes suspicion on the part of the handler, which evokes certain behaviors that may cause the dog to alert.…I've done frame-by-frame analysis of video tapes, and it's interesting when the handler stops before the dog does. You think it might have been a cue—not necessarily intentional."

A 2011 study led by the University of California at Davis neurologist Lisa Lit, reported in the journal Animal Cognition, shows how powerful a handler's cuing of his dog can be. Lit and her colleagues had 18 handlers walk their police dogs through four rooms where they were told drug or explosive scents might be hidden but where in fact there were no target substances to be found. Each team went through each room twice, for a total of 144 sweeps, and generated 225 false alerts. The alerts occurred most frequently near markers that the handlers were told indicated the presence of scents; they were even more likely at those spots than at unmarked locations where the researchers had hidden Slim Jims and new tennis balls as distractions. "Human more than dog influences affected alert locations," Lit and her colleagues concluded. "This confirms that handler beliefs affect outcomes of scent detection dog deployments."

'Searching for Ham Sandwiches'

A few of the handlers in the Lit study admitted they had intentionally pointed their dogs to the marked locations. But for the most part they seem to have communicated their expectations subconsciously, as observers did with Clever Hans, the famous German horse who supposedly could answer arithmetic questions by tapping his hoof. Although it has been more than a century since the psychologist Oskar Pfungst demonstrated that Clever Hans was reacting to the body language of his trainer and audience, the lessons of that episode do not seem to have penetrated most police departments.

Myers and Jones say dogs should be tested in double-blind situations, where neither the handler nor the observer verifying alerts knows whether or where drugs have been hidden. But such tests are the exception rather than the rule. In weekly maintenance training, Myers says, the handler likewise should not know where the drugs are. But "typically if a cop says, 'I train the dog every week,' he's hiding things and then going around and finding the things he's hidden. Putting something out, you as the handler, then taking the dog through, you are going to seriously skew the training; you're going to cue. You can't help it; you know exactly where the damned thing is."

Even when a handler does not know exactly where the drugs are, his behavior can encourage the dog to alert regardless of what the animal actually smells. Jones says bad handling practices he commonly sees as a consultant include excessive verbal encouragement ("Go find it, boy!") and giving rewards (praise, toys, dog treats) for alerts whether or not they are accurate. In a 2011 traffic stop that resulted in a fruitless car search and provoked a lawsuit, Collinsville, Illinois, police officer Michael Reichert walked his German shepherd, Macho, around a car and claimed he alerted in the front. Neither Reichert nor the dog can be seen on the dash-cam video at that point, but Reichert can be heard repeatedly urging Macho on and praising him lavishly.

Dogs that are rewarded for unconfirmed alerts may begin responding to the wrong stimuli. A dog "might just be interested in something, which could be seen as a kind of alert by the handler," says Myers, "so he rewards him for it. And pretty soon he's going to be searching for ham sandwiches."

As that example suggests, distracting smells, such as the tennis balls and Slim Jims in the Lit study, also con­tribute to false alerts. A 13-year-old girl who was strip-searched as a result of a mistaken dog alert during a 1979 inspection of her junior high school in Highland, Indiana, apparently attracted the animal's attention because she had been playing earlier that day with her own dog, which was in heat. The Sydney Morning Herald interviewed a college student who was searched at a train station after a police dog sat down next to him; the cops found a package of dog treats in his pocket. Bob Burns remembers an incident from his years as an Air Force M.P. when a dog alerted to a locker that contained not drugs but a wastebasket with a tuna can at the bottom that an officer had hidden to avoid having to clean it for a room inspection. "The dog was just hungry," Burns says. "There was a lot of embarrassment all around."

Sometimes the right smell comes from the wrong thing. Many dogs trained to detect cocaine actually react to methyl benzoate, a volatile byproduct of black-market cocaine that is also an ingredient in perfume, solvents, and insecticide. A girl whose purse was searched due to a dog alert during a 1978 sweep of her high school in Goose Creek, South Carolina, turned out to be carrying a small bottle of perfume. Similarly, acetic acid, which is what dogs smell when they smell heroin, is found in vinegar, various food products, and some kinds of glue; the same odor can be emitted by prescription drugs when they are exposed to air. Piperonal, a smell that dogs associate with MDMA, is used in artificial flavors, perfume, and mosquito repellant. Dogs also may have trouble distinguishing the smell of marijuana from the odors of fir and juniper trees.

Given all these potential sources of error, how does a judge know when a dog's alert is reliable enough to justify a search? In Harris' case, the Florida Supreme Court concluded that the search of his truck was illegal because the evidence presented to demonstrate Aldo's reliability was inadequate. The court wanted more information about Aldo's training and certification—an important issue because Florida, like most states, has no uniform standards for drug-detecting dogs. The court also wanted more details about Aldo's performance on tests ("really good," according to Wheetley). And it wanted to know his record in the field, a question Wheetley could not answer because he does not keep track of erroneous alerts. After all, who would be interested in such information?

'A Search Warrant on a Leash'

Challenging the reversal of Harris' conviction before the U.S. Supreme Court, the state of Florida (joined by the Obama administration) argued, in effect, that judges should automatically accept a police dog as reliable. "The handlers themselves are going to be in the best position to know the dogs and evaluate their reliability," Gregory Garre, the lawyer representing Florida, told the Court in October, "and they have a strong incentive to ensure the dogs are reliable." So according to Garre, if a cop trying to justify a search vouches for the reliability of a dog whose alert supposedly justifies that search, there is no reason to question him.

Garre argued that "the most important thing" in judging a dog's reliability "is successful completion of proficiency testing." How does a judge know a dog has successfully completed proficiency testing? Because the police say so. When training is done by "actual police departments," Garre said, "this Court ordinarily would presume regularity." And what constitutes "regularity" when there are no uniform standards? "We would ask whether or not the dog successfully completed training by a bona fide organization," Garre said. "We don't think it's an appropriate role for the court to delve into the contours of the training.…You would have to accept it…on its face."

And why wouldn't you? After all, Justice Antonin Scalia observed, "if the reasonableness of a search depended upon some evidence given by a medical doctor, the court would not go back and examine how well that doctor was trained at Harvard Medical School." Then again, Harvard Medical School, unlike a police department's dog training program, is accredited, based on uniform national criteria, by the American Association of Medical Colleges, and its graduates must satisfy objective, transparent tests to be licensed and certified in their specialties. Furthermore, unlike police dogs, doctors can talk, which means they can testify and be cross-examined regarding their qualifications and the reasons for their conclusions.

Scalia seemed genuinely flabbergasted not only by the idea that a dog might be inadequately trained but also by the suggestion that police might exaggerate a dog's reliability. "Why would a police department want to use an incompetent dog?" he asked Glen Gifford, the assistant public defender representing Harris. "What incentive is there for a police department?" Gifford patiently explained that "the incentive is to acquire probable cause to search when it wouldn't otherwise be available."

It should be obvious why a police officer might value a dog that alerts promiscuously, giving him license to search anyone he deems suspicious. "It's a search warrant on a leash," Myers says. "It's such an enormous back-door entry into search and seizure without a warrant."

A brief filed by the Institute for Justice in Harris highlights another motive: If a dog's alert justifies a search, it can also justify seizure of property allegedly tainted by illegal drugs. "There are countless examples of police seizing large sums of cash based on nothing more than a positive dog alert," the brief notes, even though contamination of currency with cocaine and other drugs appears to be pervasive. Since police departments typically share the proceeds from civil forfeiture, they have a direct financial interest in dogs that facilitate it.

Yet few of the justices seemed inclined to elaborate on the distinction between "a well-trained narcotics detection dog," entrusted with the power to authorize searches and seizures, and any old dog grabbed from the pound by a police department and presented as such. Jeff Weiner, the Florida defense attorney, says, "I only hope the Court will realize how incredibly naïve they have been and how they have given law enforcement a green light to do away with the Fourth Amendment merely by uttering the magic words, 'My dog alerted.'?"

Judges around the country commonly accept a dog's alert, by itself, as sufficient basis for a search, but the Supreme Court has never explicitly said it is, although passing comments in a couple of decisions can be read that way. Furthermore, the Court has always resisted precisely defining probable cause, the standard for issuing a warrant (or for upholding a car search, which can be conducted without a warrant but is supposed to meet the same test). Probable cause, the Court has said, means there is "a fair probability" that evidence of a crime will be discovered. It is not clear how reliable a dog must be to satisfy that standard. A 4 percent chance of finding contraband based on a dog's alert, as in the Florida roadblock operation, presumably would be too low. What about a 20 percent chance, as in the latest data from New South Wales, or a 44 percent chance, as in the Chicago Tribune study?

"Who determines when a dog's reliability in alerting has reached a critical failure number?" Justice Sonia Sotomayor asked during the oral arguments in Harris. "I'm deeply troubled by a dog that [accurately] alerts only 12 percent of the time.…That seems like less than probability." Gifford observed that "in the lower courts, once you get below 50 percent, probable cause is much less likely to be found." The 2006 Australian study found that the accuracy of 17 police dogs used to sniff out drugs on people ranged from 7 percent to 56 percent. This wide variation underlines the importance of assessing the ability of each dog-and-handler team on an individual basis, rather than accepting blanket assurances that all dogs and handlers have been properly trained.

'Franky's Nose Is Not Technology'

While the Supreme Court seemed reluctant to require greater skepticism of such claims, the justices were more receptive to concerns about using dogs to identify homes containing drugs. In fact, Scalia, the justice who was most clearly hostile to questions about police dogs' professional credentials in Florida v. Harris, was the one who was most indignant about bringing them to people's doorsteps without a warrant in Florida v. Jardines.

On the morning of December 6, 2006, based on an unverified Crime Stoppers tip received a month earlier, Miami-Dade Police Detectives William Pedraja and Douglas Bartlet approached the Princeton, Florida, home of Joelis Jardines, where the tipster had said marijuana was growing. They were accompanied by several DEA agents and Franky, a chocolate Labrador retriever. Bartlet brought Franky up to the entrance of the house, where he sniffed around for a minute or two before sitting down at the front door. After Bartlet announced that Franky had alerted to the house, Pedraja approached the front door and claimed he could smell marijuana, although Bartlet said he did not.

Based on Franky's alert, Pedraja obtained a search warrant that police executed later that day, finding 179 marijuana plants, growing equipment, and Jardines escaping out the back door. Charged with trafficking in more than 25 pounds of cannabis, a first-degree felony punishable by up to 30 years in prison, Jardines successfully argued that the evidence against him should be suppressed. The Florida Supreme Court concluded that the search of Jardines' home was illegal because Franky's inspection of the area near the front door was itself an illegal search.

Since the U.S. Supreme Court has said that using a dog to check luggage at the airport or a car during a traffic stop does not count as a search, you might think that Franky's sniffing at Jardines' doorstep would not qualify as a search either. But during oral arguments the justices seemed inclined to agree that homes are different. Jardines' lawyer, Howard Blumberg, argued that "when a police officer goes up to the front door with a narcotics detection dog" he has "physically trespassed, because there is no consent to do that, onto a constitutionally protected area, the curtilage of the home"—that is, the area immediately surrounding it. That argument appeared to be crafted with Scalia in mind. In U.S. v. Jones, the 2012 decision in which the Court ruled that tracking a vehicle by attaching a GPS device to it requires a warrant, Scalia's majority opinion emphasized the physical trespass required to install the device. "If you…follow the test set forth in Jones and apply it to what happened here," Blumberg said, "it is a trespass."

Scalia signaled that he was receptive to this approach even before Blumberg got up to speak. "Police are entitled to use binoculars to look into [a] house if the residents leave the blinds open," he told Gregory Garre, who represented Florida in this case as well as Harris, but "they're not entitled to go onto the curtilage of the house, inside the gate, and use the binoculars from that vantage point.…Why isn't it the same thing with the dog?…It seems to me crucial that this officer went onto the portion of the house…as to which there is privacy and used a means of discerning what was in the house that should not have been available in that space.…Police officers can come there to knock on the door…[but] when the purpose of the officer's going there is to conduct a search, it's not permitted…He's going there to search, and he shouldn't be on the curtilage to search."

Scalia also wrote the majority opinion in Kyllo v. U.S., the 2001 case in which the Court held that using a thermal imager to measure the heat radiating from a home (as evidence of grow lamps) requires a warrant. Justice Ruth Bader Ginsburg read Garre a passage from that decision, which she suggested applies to drug-sniffing dogs as well: "We think that obtaining by sense-enhancing technology any information regarding the interior of the home that could not otherwise have been obtained without physical intrusion into a constitutionally protected area constitutes a search, at least where, as here, the technology in question is not in general public use."

Garre responded by arguing that "Franky's nose is not technology," since he is merely "availing himself of God-given senses." But in its natural state, Franky's nose does not tell police when molecules of certain chemical compounds are floating through the air; that requires human intervention, based on technical knowledge, aimed at turning a descendant of wolves into a law enforcement tool. "The dog per se is not a technology," says Myers, but "the dog is part of a technology that has been applied to a particular use." The main point, when it comes to expectations of privacy, is that Franky, like a thermal imager, enables police to find evidence they could not detect with their own unaided senses. They can thereby obtain information about what is happening inside a home that they otherwise could get only by entering it.

'Dogs Are Not Magic'

But dogs do not perform this function inerrantly. The notion that a dog sniff is not a search and the notion that a dog sniff justifies a search are both based on overblown notions of canine capabilities, a fact that makes the implications of those ideas all the more troubling. A cop already has the authority to stop cars for minor (and possibly imagined or invented) traffic violations that people routinely commit. If you give him a dog he can deploy during any stop to justify a search, a dog whose alerts may be imagined, invented, or triggered by deliberate or subconscious cues, he now has the ability to search cars at will.

In Caballes, the decision that gave police this ability, Justice Souter warned that "an uncritical adherence to Place [which held that a sniff is not a search] would render the Fourth Amendment indifferent to suspicionless and indiscriminate sweeps of cars in parking garages and pedestrians on sidewalks." During the oral arguments in Harris and Jardines, Justices Sotomayor and Ginsburg likewise worried aloud about police taking dogs from door to door in an apartment building or from house to house on a street. Garre, Florida's lawyer, argued that limited resources and "community hostility" would discourage such operations.

But they are already happening. In 2011, for instance, The Roanoke Times reported that police in Pulaski, Virginia, had been using dogs to randomly search for drugs in apartment complexes "for a couple of years." Last spring the Fargo Housing Authority in North Dakota announced plans for similar sweeps. Students as young as 6 have been randomly subjected to dog sniffs at public schools throughout the country for decades.

Such olfactory dragnets would be disturbing enough with dogs that are 100 percent accurate. But with actual dogs, which could be wrong most of the time or even nine times out of 10, they are little more than pretexts for police to search wherever and whenever they please.

"Dogs can be a very good and useful tool," Myers says. "But people have gotten both lazy and superstitious about that use. Dogs are not magic."