The Sniff Test

Machines might be able to detect chemicals or explosives in suspicious bags, but dogs also can go anywhere, give comfort or intimidate, and learn.

The subway system in Philadelphia was having a problem with suspicious packages. Each unattended backpack or bag that could not be confirmed as unthreatening occasioned a call to the bomb squad-essentially shutting down a station, sometimes for hours at a time.

The Southeastern Pennsylvania Transportation Authority had bomb-sniffing dogs, but they sometimes zeroed in on the odor of explosive residue on benign items containing the same chemicals, such as the grease on subway doors. After the transit bombings in Madrid and London, no one wanted to handle an unattended bag without knowing for sure what was inside.

So in February, SEPTA bought two suitcase-sized explosive detection units from Irvine, Calif.-based HiEnergy Technologies. The detectors, which cost $300,000 apiece, use neutron-based technology to scan an item and tell officers with 97 percent accuracy whether explosives are inside, and if so, what quantity. The new technology seems fast, convenient and certain. Yet SEPTA officers still swear by the centuries-old method: the nose of a four-legged companion. The new detection units were meant to supplement, not replace, bomb-sniffing canines, they say.

"I don't think you can replace dogs," says Capt. Jack Wenke of SEPTA's Transit Police Department. "You certainly need to supplement them. [But] there's always going to be a need for dogs." Even as technology becomes more and more sophisticated, it cannot push nature's oldest detector out of the security market. In fact, most security officials-including many working for the firms building new detectors-insist that bomb-sniffing dogs never will become obsolete. Competition between proponents of canine- and machine-based detection has been supplanted by cooperation. "I don't think there will ever be a situation where dogs are considered less necessary," says Ross Harper, lead research scientist in national security and homeland defense for ICx Nomadics, an Oklahoma City manufacturer of weapons detectors.

Machines Can't Walk

Sept. 11 and subsequent attacks and bombing plots have spotlighted the successes and shortcomings of explosive detection machines, from the minivan sized portals used to scan high volumes of luggage to hand-held devices used by officers on the border, at seaports and elsewhere. The Transportation Security Administration has more than doubled the number of dogs on its National Explosives Detection Canine Team since Sept. 11. The agency now has 425 bomb-sniffing dogs at 85 airports and 11 mass transit facilities across the country, and officials say that number will probably reach 500 by the end of the year.

To be sure, that figure is far below the more than 6,000 explosive trace detectors at airports nationwide. Still, officials with TSA's canine program say plans call for the dog team to continue expanding until at least 2011. Many report that the demand for dogs now outpaces the ability of TSA and other training organizations to provide them. "[The demand] continues to go up," says Dave Kontny, director of TSA's canine program. "Although we're five years away from 9/11, we're only one year away from London."

Bomb-sniffing dogs essentially work like breathing, tail-wagging versions of electronic trace detectors seen most often in airports, which swab a piece of luggage and analyze it for chemicals that could indicate explosive use. Explosive material is extremely sticky so tiny bits of residue tend to remain on the body and materials of those who have handled it. A dog's nose is so powerful that it can sense even microscopic amounts of such chemicals: Dogs have about 220 million olfactory cells that can respond to an odor compared with about 50 for a human, and each of those canine olfactory cells has between 100 and 150 tiny nerve endings called filia that further aid smell. (Human olfactory cells have about six to eight filia).

There's no definitive comparison be-tween the sensitivity of dogs and detection machines, but most experts say the canine nose is as or more acute than its mechanized counterpart. This prodigious sense of smell has led to the use of dogs to detect termites in houses, mold in walls and even cancerous cells in the human body, in addition to their more famous ability to sniff out concealed humans and narcotics.

"No one has really been able to measure how a dog does what they do," says Scott Thomas, the breeder for the TSA program. "We know what they do. We know they do it accurately."

The Bureau of Alcohol, Tobacco, Firearms and Explosives, the Coast Guard, the Homeland Security Department's Customs and Border Protection bureau, private organizations and many state and local law enforcement agencies use bomb-sniffing dogs. Before Sept. 11, dogs were dispatched mostly to find narcotics. The program run by the Federal Aviation Administration before TSA's creation in 2001 deployed canine teams at strategic airports around the country. If a plane faced a bomb threat, it was to divert to one of those airports. The terrorist attacks changed that. TSA expanded the program to place teams at all of the nation's largest airports and shifted training toward explosives detection.

To learn to detect explosives, dogs must undergo months of training on any of the various chemical compounds that are present in explosives. For example, many explosives are based on nitrates. (Nitrates also are present in fertilizer, cleaning solutions and tennis balls, which can cause false alarms.) TSA is the second largest government canine training organization-the Defense Department is the largest-and teaches dogs at a large facility at Lackland Air Force Base in San Antonio, Texas.

Shared with the Defense Military Working Dog School, TSA's San Antonio facilities resemble real mass transit stations to teach dogs to function in crowded public spaces: a mock airplane fuselage, a faux subway station with two authentic New York City transit cars, and a replica of a terminal from Atlanta's Hartsfield-Jackson International Airport, whose ticket counters and photographic mural backgrounds make it look nearly identical to the actual terminal.

The initial 100 days of training focus on getting the Labrador retrievers and vizslas favored by TSA to associate the targeted odor with a toy. First the young dogs learn to search for a rubber treat-filled toy known by its brand name, Kong. Then they learn to target the chemical odor and are rewarded with the Kong. This reward continues throughout the dogs' work in the field; TSA officials are fond of saying that their canine detection teams are not working, but playing.

After initial training, the dog is assigned to a handler, and the pair takes a 10-week course together. Many handlers are local law enforcement officers from around the country; TSA pays about $50,000 annually per team in exchange for a guarantee that the duo will spend 80 percent of its time policing mass transit facilities. After graduation-there is a formal ceremony-the team is deployed to an airport or train station, where they have 60 days to get acclimated before a final test for official certification. (The team must be recertified each year.)

TSA officials do not publicly discuss which of the more than 19,000 potential explosive compounds the dogs are trained to find or how many each dog can detect. But they use enough training aids to ensure that teams are certified on specific threats that come directly from TSA's intelligence division, they say. This speaks to one of the most commonly cited advantages of canine detection over machine-based detection: The increasing use of improvised explosives means constantly updating detection for new chemicals, and most detection pros say updating is easier with canines. Kontny says the TSA program can add a compound to a certified dog's detection repertoire within days or weeks, and did so earlier this year, when British authorities arrested dozens of suspects allegedly planning to bring down transatlantic flights with liquid peroxide-based explosives. "If we determine a new threat, in most cases it's a little easier to train that dog on a new explosive odor than it may be to recalibrate [a machine]," Kontny says.

There are other advantages. For one, machines can't walk. While large, stationery detectors are relegated to checkpoints, dogs can be set loose to search a plane on the tarmac or an entire terminal for explosives. "Right now, I don't think anybody could argue with the portability that dogs bring to the table," says John Dowd, a sales engineer for Isonics Homeland Security and Defense Corp., a Columbia, Md., manufacturer of hand-held chemical and explosive detectors. "Their ability to get into areas that . . . we haven't even ever been able to take any kind of detector into has made them absolutely invaluable."

And then there's the intangible, psychological effect a dog has in comforting passengers and deterring potential terrorists, qualities that lifeless, inconspicuous machines lack. "Increased visibility of explosive detection dogs together with surveillance cameras and a perceptible law enforcement presence may well make a terrorist think twice," said Rep. Bennie Thompson, D-Miss., at a September 2005 hearing before the House Homeland Security Subcommittee on Management and Oversight.

Horny, Hungry and Tired

Of course, detection dogs bring their own set of drawbacks. Though TSA has been roundly criticized for the soaring maintenance costs of its electronic detectors, the dogs require a large amount of maintenance, too. "What we say in industry is they get horny, hungry and tired," says Peter Kant, vice president of government affairs for Rapiscan Systems, a Hawthorne, Calif., firm that markets various detectors.

Dogs need breaks. They can't work around the clock in times of heightened alert. And the longer they work, the less accurate they become. A study by Auburn University's Canine Detection Training Center calculated a 12.5 percent false alarm rate generally, but found that the error rate increased the longer the dog searched without a break. When on duty for between two and three hours, the false alarm rate for dogs in the study was 60 percent. Harper calls this a calibration problem. "You can tell instantly if an instrument is having a bad day, because the first thing you do is a [diagnostic test]," says Harper, of ICx Nomadics. "When you open the back of the car and let the canine out, it's much harder to tell if the canine is on, primed and good to go that day."

Theoretically, reading a dog is the handler's job. If a dog is sick, tired or just generally less than excited, it's the handler's responsibility to realize that the dog might not be catching everything. That, and noticing when the dog sniffs something suspicious, is easier said than done. "The dog doesn't stop and stand up on his two legs and say 'Hey, I found a block of TNT,' " says Rick Stauber, a retired master explosives ordnance disposal technician for the Army and now a consultant with Plexus Scientific Corp. in Alexandria, Va. "No matter how good the dog handler and dog folks think the dogs are, you don't have that dog talking to you directly."

Detection machines, on the other hand, can do just that. Wary of systems that rely on interpretation by checkpoint workers and security agents, technology firms have been building more automation into their devices. The neuron-based scanning of HiEnergy's SIEGMA 3E3 and others claim to go beyond alerting operators to potentially dangerous materials by also identifying chemical compounds. "You're getting the doglike detection along with the benefits of imaging and of automation," Kant says. "As these things come online, the requirements for dogs at airports become less."

But most observers, even on the business side, say technology will never replace dogs. "To rely on any one element, whether a biological solution or a technological solution, is a huge error," says Thomas. "The fact that we can send luggage through a sensor and not have a dog screen every bag that comes through the airport makes good economic sense. At the same time, if an airplane runs out on the tarmac [with a bomb threat], we have a tool we can take right out there immediately-and a tool that has not been calibrated to one or two odors, but whatever odors we have identified as likely threats."

Designer Dogs

In fact, far from rendering dogs obsolete, Thomas says technology has made them more valuable. Scientists have decoded the dog's genome-the sequencing of its DNA-meaning they have identified the set of genes that shape specific behaviors, such as "independent possession," a dog's desire to get a toy and hang on to it. This means breeders can target specific characteristics that make for better bomb-sniffers. "We'll be custom-designing dogs for purposes of detection," says Thomas. "We call them Labrador retrievers; there may come a day we call them Labrador detectors." Thomas says TSA received a $2 million grant for genetic research to further define which genes produce instincts especially suitable for bomb-sniffing.

And the agency already has begun such work. Many of its dogs come from Europe, where most other U.S. government agencies and private organizations purchase their detection dogs. TSA has a breeding program, which began at the start of this decade and now produces 100 puppies a year-each named after a victim of Sept. 11. The program includes Labrador retrievers, because of their instinct for toy play; vizslas, because they are more sensitive smellers; and now, a new custom breed: the Vizslador. By crossbreeding, TSA hopes to produce puppies with the best attributes of each. In the future, Thomas says the agency is interested in incorporating the signaling instincts of the "pointing breeds," which signal hunters when they smell game. "Let's not think of a dog as an old tool that can't be improved on," he says. "It can, with current technology."