by Ken Chiacchia, PhD, WEMT-B

Recently, my SAR dog team began training for scent-discrimination work -- searching for individual human beings based on their scent. After more than 10 years of handling an air-scenting dog who searches only for generic human scent, the experience has been an education.

Watching a poorly trained discrimination dog work can be painful; watching a well-trained dog is transformational. Before my eyes my dog has become amazingly adept at picking out the "right" person's ground trail from the trails of other people. We're not ready to hang up a shingle for this service yet -- but we're getting there.

Alongside learning how to work with my dog as a scent-discrimination team, I've been learning about what scientists know about individual scent. The discoveries of the last 10 years have turned what we thought we knew about individual scent on its head. Rather than an almost accidental byproduct of bodily waste, diet, and other environmental factors, it turns out that the body has a very purposeful system for generating and broadcasting individual scent.

Human scent: Old school

The old wisdom was that while the body's many types of sweat glands all contribute to our smell, the sebaceous and apocrine glands are the most important. The sebaceous glands secrete an oily substance containing chemicals that bacteria break down into short carboxylic acids that have distinctive, "sour" smells.

Concentrated in the armpits and the genital region -- but mostly in the armpits -- the apocrine glands produce chemicals called odiferous steroids. These either possess a musky smell or can be broken down by bacteria into smelly compounds.The parts of the body where apocrine glands reside tend to have wiry hair (which may help diffuse smells) and a unique population of bacteria, according to Claus Wedekind, an evolutionary biologist at the University of Edinburgh, Scotland. When we apply deodorants -- which are actually bacteria-killing antibiotics -- to the armpits we reduce bodily smell.Besides the contribution of the sebaceous and apocrine glands to human scent, other obvious components of scent are the many foreign substances that hitch a ride on our bodies. Hitchhikers may include traces of insect repellants, fossil fuels, food residue, perfumes, and so forth. Washing probably knocks these hitchhiker smells down somewhat, but only adds soap smells.

Another wrinkle to the description of scent is that the odor coming off our bodies has two components -- odiferous gases and microscopic "rafts" consisting of small numbers of or single dead skin cells. For the purposes of this discussion, we can regard skin rafts as miniature generators of scent much like the rest of the skin: while they aren't scent themselves, they contain all the substances that the skin does and generate odiferous gases in the same way. These rafts probably constitute an important part of the ground scent left behind a person, as they continue to generate scent as long as their cargo of odor-generating substances lasts.

There are a couple of problems with assuming that this classic description of human scent can fully explain individual, or even generic, human scent. First, while it's easy to prove what causes human body smells that humans perceive, there's no guarantee that they are what the dogs are using to identify us. Second, the sebaceous and apocrine glands don't become active before puberty, and SAR dogs don't seem to have trouble finding children (though some handlers have reported atypical reactions the first time their dogs find one, so these glands could play a role).

More importantly, in the last decade a number of laboratory findings have come together to suggest that this description of human scent misses the point, at least partly.

Of mice and men: Smell and the MHC genes

The new discoveries were built on lab findings after World War II, when researchers began to tease apart why animals reject transplanted organs from one donor but not another. Molecules called the major histocompatibility complex (MHC) proteins, the scientists found, sit on the surface of all the cells in the body. The MHC proteins are, in effect, the body's bar-code system: except for identical twins, each individual has a unique combination of the six MHC genes that direct the production of unique MHC proteins.

In the 1960s and 70s, researchers began to breed strains of mice that were genetically identical except for their MHC genes, and used these bloodlines to pick apart exactly how the MHC proteins work. Normally, the MHC proteins display fragments of other proteins that are present inside the cell. Much like a state trooper making a sobriety check, infection-fighting white blood cells then come by and check each cell's MHC proteins. If all the white blood cell sees are familiar MHC proteins displaying normal cell fragments, nothing much happens.

If, on the other hand, the MHC proteins are displaying alien fragments -- such as those from bacteria or viruses -- the white blood cell will kill the other cell to stop the infection from spreading. If the cell has MHC proteins that don't match the white blood cell's -- such as you might find on a transplanted organ -- the white blood cell kills that cell as well, mistaking it for an infected cell. This kind of white-blood-cell attack underlies rejection of transplanted organs.

According to Gary Beauchamp, a scent researcher at the Monell Chemical Senses Center in Philadelphia, in the 1970s the famous immunologist Lewis Thomas began to noodle with the idea that every cell in our bodies carries the body's molecular "fingerprint." Was it possible, Lewis wondered, that animals used this system to identify each other by smell as well? The idea "was purely speculative, off the wall," says Beauchamp.

It was also dead-on correct.

Female mice, researchers discovered, will mate more often with males whose MHC genes differ from theirs than those with similar MHC genes. Somehow, the different MHC genes were giving males different smells that the females were picking up.

Recently, molecular biologist Lara Carroll of the Howard Hughes Medical Institute at the University of Utah found that the female mice could sniff out a difference of only five amino acids out of the hundreds that make up the MHC proteins. This difference is much smaller than wild mice would have with each other, and so the system generates more than enough individual smell to identify individuals.But how can mice smell MHC proteins inside the body?

A smelly consequence of immunity? Individual scent and MHC breakdown

It wasn't until the 1990s that Beauchamp and colleagues began to unravel how individual MHC proteins may lead to individual scents. They began by training mice to identify the urine of other mice, much as a SAR dog handler may train a dog to identify people by the scents of their hats.

Beauchamp and his team split the mouse urine into its chemical components, using their "detection mice" to indicate which parts still carried the individual scent. They discovered a bouquet of chemicals, mostly carboxylic acids, that seemed to convey individual scent.

More interesting, they found out that the mice could also recognize each other by sniffing each other's blood serum -- recall that white blood cells are the killers of MHC-mismatched cells. The only trick was that you needed to chemically break down the MHC proteins in the blood first: As long as the proteins were intact, there was no individual smell.

The discovery began to explain how the body broadcasts its "scent identity." As part of their normal function, MHC proteins stick to and display fragments of other proteins; while these fragments normally don't have a smell, one possibility is that they are broken down into smaller, odor-carrying molecules. On the other hand, it could be that the MHC proteins pick up smelly substances in the blood as well as protein fragments.Either way, when blood serum is processed into urine in the kidneys, the body breaks down the MHC proteins, releasing free odorants into the urine. Untreated blood serum has no individual smell because the odorants are still stuck to the MHC proteins. But digest those proteins chemically, and these odorants are free to find the smeller's nose.  (See the Figures.)

 Figure 1: MHC generation of an immune-systrem attack.

 Figure 2: MHC generation of individual odor.

Even better, the explanation fits dog handlers' experience perfectly. Since each person has an individual environment with respect to food, perfumes, contaminants, and so forth, each person's bloodstream contains a unique mix of potential odorants for the MHC proteins to capture. Since each person also has an individual set of MHC genes, his or her MHC proteins will select a different mix of those potential odorants to stick to. Both genetics and environment play an important role.

The MHC angle also nicely explains research showing that dogs can sometimes discriminate between twins -- but not always. You'll recall that identical twins share MHC identical MHC genes; but if they are eating, breathing, and absorbing different things, their identical MHC proteins will nevertheless be picking the scent bouquet from a different pallet of potential odorants: nature and nurture.

"You can't really distinguish between the two," says Carroll. "The MHC binds odorants in its environment, so MHC plus environment is what produces an individual odor -- that's really important."

The human connection

But how applicable is this research -- done, after all, with rodents smelling other rodents -- to dogs smelling human beings?

The entire MHC system is definitely present in humans: The human version of the MHC proteins, called human leukocyte antigen (HLA) proteins, are part of how doctors match organ donors with recipients to avoid rejection. In addition, research by Wedekind and others has shown that people can "smell" each others' HLA genes.

"T-shirt odors were judged as more pleasant when they were worn by men or women whose MHC genotype was different from that of the smeller, a finding that is analogous [to] the one in mice," Wedekind wrote in an article that appeared in Current Problems in Dermatology in 2002.

Studies by other researchers showed that women can reliably identify by smell clothing worn by their own babies among clothing worn by others, and that humans tend to select mates with dissimilar HLA genes, just as mice select mates with different MHC genes. Scientists aren't sure why either mice or humans prefer MHC-dissimilar mates. It may help reduce inbreeding, or it may "mix and match" MHC diversity, making the species as a whole more resistant to new infections.Humans may not have anywhere near dogs' ability to discriminate fine differences in individual scent -- but we have some ability, and it seems to affect how we relate to each other. "Animal magnetism" between people, after all, may have more to do with HLA-generated smells that many of us would like to think.

Other investigators have shown that traces of human HLA proteins can be found in armpit sweat, and that differences in odorants found in this sweat parallel HLA differences, just as in mice. So most researchers agree that this individual scent broadcasting system exists in humans.

In addition, all the signs are that dogs have the olfactory receivers to detect this broadcast, both from each other and from humans. "I'd be completely dumbfounded if it weren't true for dogs," says Beauchamp.

Practical lessons for SAR dog handlers

All of which brings us back to the question of scent discrimination in SAR dog work. Understanding how scent is generated can lead to practical applications for scent-discriminating dog teams. In addition, it contributes to a level of sophistication that invariably improves our performance in the long run. However, we need to choose these applications carefully. A misapplied or misunderstood application of a laboratory result -- which, by definition, can never be the "last word" on the subject -- can cause us to be less effective, adding risk to our search subjects.

What the scientists have discovered about individual scent hasn't called into question most practices in training scent-discrimination dogs, although it does suggest that armpit and genital smells may be particularly important. One method we've run across, of having practice subjects drop cotton balls from their hip pockets to help dogs focus on the trail, may rely on scent from the genital region diffusing through those pockets.

In fact, one research study showed that dogs can have trouble matching the scent on a person's hand with that in the crook of the elbows. Again, if the name of the game is detecting traces of MHC smell from the armpits or genitals, comparing isolated parts of the body like that could indeed trip many dogs up. For this reason, the common practice of mixing up scent articles between items like cotton balls, car seats, keys, hats, other types of clothing, and even blood is probably particularly wise.

One fascinating twist of the MHC-smell story is that the body may have the ability to generate smell on its own, without the help of bacteria -- recall how the kidney may liberate MHC odorants in making urine. This is a bit of dynamite for many dog handlers. Since dog handler William Syrotuck wrote about scent in the 1970s, most of us have literally grown up thinking of bacteria as a vital part of human scent.

Surely, the conditions that favor human scent generation in the field -- moisture and heat -- can be explained either because they also happen to favor bacterial growth or that they may favor slow and steady release of MHC-produced odorants. But the idea that bacteria aren't necessary suggests there may be conditions that favor scent but don't favor bacteria. Dog handlers may want to keep that point in mind as they train and practice.

It's difficult to tell how MHC scent would affect the "ageing" of a trail. Most trailing handlers agree that scent trails change both in character and intensity with time, and that at a certain point they exhaust themselves. This would be true whether the scent depends on bacteria generating scent from digesting skin raft components or on MHC fragments carried by the raft slowly releasing odorants, though, and it's not yet possible to say which would "run out" more quickly. Certainly, dog handlers' experience in the field is a more reliable indication of this than any theoretical analysis of how the scent is generated.

The most relevant experiments to this question to date -- done in mice and rats -- unfortunately offer a frustrating contradiction. Mice can be raised in a germ-free environment that isolates them from any bacteria, and such mice have an individual scent. On the other hand, rats seem to lose their individual scent when raised germ-free. So this question is still very much open; it's certainly possible that both the MHC system and bacteria work together to produce the scent our dogs detect in the field.

There is one very important practical lesson from the realization that bacteria may play less of a role -- if any -- in generating scent than we thought. I've heard the suggestion repeated that, because bacteria from a trail or other non-living scent source are unlikely to grow in sub-freezing temperatures, dogs may not be effective at finding these scent sources in such conditions -- and certainly, such temperatures do pose challenges to scent work.

If bacteria aren't playing the role we thought, though, we need to re-examine this assumption. It may be that there is a far less abrupt effect of temperature in these cases, and that running, for example, a trailing dog on a sub-zero day might save a life. While we have never heard of anyone withholding the dog resource due to a mistaken belief in the role of bacteria, its possibility underlines the importance of not taking scientific theories too far when lives are at stake. 

*This article appeared in Advanced Rescue Technology,  Vol. 7 No. 3 pp. 45-50. Copyright 2004 by Summer Communications, Inc. Reproduced with the permission of the publisher.

Further reading

General information on scent

Scientific reviews on individual scent

Brown, J.L. and Eklund, A., "Kin recognition and the major histocompatibility complex: An integrative review," The American Naturalist 143(3):435-61, 1994.

Ferstl, R., et al., "MHC-related odors in humans," in Doty, R. L. and Müller-Schwarze D. (eds.), Chemical Signals in Vertebrates VI, Plenum Pres, New York, 1992.

Wedekind, C., " 'Good' and 'bad' body odours," pp. 23-29 in Kreyden, O. P. et al. (eds.), Current Problems in Dermatology. Vol. 30: Hyperhidrosis and botulinum toxin in dermatology, Karger Verlag, Basel, 2002.