Whether they waft from a crackling campfire or a rose in bloom, odor molecules are volatile: they readily vaporize either from heat or simply from exposure to the air. Most smells are a combination of these particles. Some we inhale through our nostrils; others we take in when chewing, which opens a channel connecting the throat and nose. (This is the passage to the “all-powerful joy” Marcel Proust feels when he eats a tea-soaked madeleine in Remembrance of Things Past.)

We perceive smells with olfactory sensory neurons, specialized cells found high inside our nasal cavity. When an odor molecule connects with a receptor on an olfactory neuron, the neuron sends a signal to a structure at the front of the brain called the olfactory bulb. This is the first relay station for processing smells.

When anatomists studied human and animal brains in the 19th and early 20th centuries, they saw that, compared to other mammals, the human olfactory bulb was proportionately smaller and positioned differently, tucked under the frontal lobe instead of right up front. To them, this implied that smell was relatively unimportant for humans. French surgeon Paul Broca devised an influential theory that free will rested in the frontal lobe, the largest part of the human brain. As humans developed the ability to reason, Broca argued, their olfactory bulbs shrank to make room, and what he called the “bestial” sense of smell withered. (Broca’s work—including his belief that the olfactory bulb was especially diminished in “men of the white race” compared to men of “inferior races”—helped lay the foundation of eugenics.)

For the most part, the notion that humans had an atrophied sense of smell remained unscrutinized until well into the 20th century. The topic started to attract new attention in the 1960s and 1970s, mainly from researchers trying to develop a general theory that connected chemical structure to odor. A major breakthrough came in 1991 when Columbia University scientists Linda Buck and Richard Axel mapped the genes, first in rats and subsequently in mice, that create olfactory receptors. Buck and Axel received the 2004 Nobel Prize in Physiology or Medicine for this work and subsequent studies showing all mammals had large families of genes that coded for large numbers of odorant receptors. In one of the newest findings in this area, a pair of studies published in 2026 showed that scent receptors in the noses of mice occupy specific, predictable positions—what one of the researchers called a “long lost map for smell.”

Today, scientists estimate that humans can distinguish at least one trillion different smells. And it doesn’t take much material to do so. Many people can detect mercaptans, for example, at air concentrations of just a few parts per billion—equivalent to a few drops of liquid in a 10,000-gallon municipal swimming pool.

What’s more, our sense of smell can improve with training. In a 2006 experiment, researchers at the University of California, Berkeley, blindfolded 32 volunteers and asked them to follow a 10-meter string scented with chocolate through an open grass field, working solely by smell. Two-thirds of participants succeeded. When several repeated the challenge over multiple days, they doubled their speed. They spontaneously imitated dogs, zigzagging along the scent trail and sniffing faster as they accelerated.

Other research shows that the human olfactory bulb responds differently to smells depending on whether it identifies them as pleasant or unpleasant. Negative odors are processed earlier—within as little as 300 milliseconds—and trigger a physical response that causes a person to lean away from the source. In a study in which young men sniffed either strong or weak concentrations of rosemary oil (a pleasant smell) or ammonia (unpleasant), the ammonia produced a larger alerting response in their brains, regardless of its concentration.

These findings suggest that our sense of smell is designed to help us detect potential threats in our environment—and not just gas leaks. From spoiled foods to strange odors creeping out from under the hood of your car, bad odors have the power to tell us, “Something’s wrong here.”

Nevertheless, warning systems rely far more often on our eyesight than on odors. And the notion persists in popular culture that smell is humans’ weakest, and by implication, least important sense. In a 2021 survey of college students and adults, for example, nearly half of women surveyed were willing to give up their sense of smell to keep their hair; one-quarter of the college students would give up smell to keep their phones.

Smell’s underappreciation may come down to its complexity. Vision is comparatively straightforward. The colors we see are produced by just three types of receptors in our retinas detecting light at different wavelengths—short (blue), medium (green), and long (red). In contrast, scientists estimate there are about 400 classes of olfactory receptor cells in the backs of our noses. Each olfactory cell can detect more than one smell, and a single odorant can trigger more than one type of receptor, although it may fit some of them better than others. This intricate system has frustrated scientists’ efforts to classify smells based on objective criteria, such as molecular structure.

“There’s not a one-to-one relationship between some set of chemical parameters and quality perceptions,” says Richard Doty, a professor of medicine at the University of Pennsylvania and director of the private Smell & Taste Evaluation Center. Human response to smells, he asserts, “is all context dependent, and it’s largely learned.”

What we smell is influenced by genetics, culture, expectations, and experience. We are taught to think of many smells as either good or bad, and our perceptions are influenced by context. If you ask someone to sniff a concealed object and tell them it’s an expensive goat cheese, they’re more likely to at least feign approval than if you tell them it’s a dirty sweat sock. (Volatile fatty acids are key sources of both of these funks.)

Associations also matter. For Robert Duvall’s Lieutenant Colonel Bill Kilgore in Apocalypse Now, the smell of napalm signals victory. In a real-life version, candlemaker Billington Farms introduced a scented candle in 2023 that smelled like burning JP-8, the main jet fuel used by the U.S. armed forces. The company had learned through surveys that for many military veterans and aviators, the smell evoked memories of excitement and adventure.

Natural gas odorization shows that such associations can be taught, regardless of whether people like or dislike a smell. But that process may take years or decades, and it has its shortcomings.

The first gases to warm homes and power street lamps came with built-in stinks. Known as town gas or coal gas, they were introduced in Britain in the late 18th century, manufactured from coal or petroleum, and contained impurities—including hydrogen sulfide, which has a strong rotten-egg odor that made leaks detectable.

The Bunsen burner, invented in 1854, changed the game. This device mixed natural gas with air before ignition in a ratio that produced complete combustion and maximum heat without generating byproducts like soot or carbon monoxide. With cleaner combustion, the association between gas and noxious odors began to fade.

Starting in the 1920s, U.S. gas producers built out a national pipeline network to deliver their product to cities and towns. As natural gas became increasingly available, it displaced manufactured gas as a fuel for home heating and cooking as well as commercial electricity production. Ads touted the new fuel as clean, healthy, and effortless to use. An American Gas Association campaign called it “the magic flame that will brighten your future.” But that cleanliness and convenience masked a growing hazard: as more odorless natural gas flowed into more homes, leaks and explosions became commonplace.

News accounts from communities across the United States in the late 1800s and early 1900s are peppered with reports of gas explosions. In November 1898, a blast in the basement of the U.S. Capitol reduced part of the Senate’s wing to rubble. There were no injuries, likely because the accident occurred on a Sunday.

Other outcomes were worse. In March 1897, an explosion from a damaged gas line next to Boston Common killed at least 10 people, injured dozens, and destroyed several streetcars. And in February 1917, a gas explosion collapsed two three-story tenement buildings in Chicago, killing at least 13 people and injuring many others.

When several explosions followed Los Angeles’s switch to natural gas in 1927, the city’s providers began testing odorants to make leaks in the system more evident. The utilities tried ethyl mercaptan, which today is used in propane, but ruled it out because of its high price and revolting smell. Instead, they turned to a product made by Standard Oil of California called Calodorant, a blend of alkyl sulfides, disulfides, and mercaptans dissolved in gasoline. Complaints of gas leaks soared, and soon other municipalities followed L.A.’s lead, including New Orleans and Denver.

Then a disastrous explosion at the New London School in rural northeast Texas in March 1937 made gas leaks a national concern. A faulty connection allowed gas to steadily collect in the school’s basement; when a shop instructor turned on a sander, the mixture ignited. The massive blast collapsed the main building, killing nearly 300 students and teachers.

Within weeks the Texas legislature voted to require odorants in natural gas to make it detectable. Other states followed suit. In 1968 Congress passed the Natural Gas Pipeline Safety Act, which required odorization nationwide.

Today millions of miles of pipeline move combustible gas across the United States—first from production sites where gas is extracted, to processors where it’s refined, then to communities. Suppliers typically add odorants at city gates, the points where long-distance pipelines feed the gas into local networks that deliver it to individual homes and businesses.

To give gases that unmissable whiff, companies select from nine commercial odorants. Their formulas vary to fit specific requirements for each type of gas and the different pressures they are subjected to during the stages of distribution. All are designed to be chemically stable in pipelines and insoluble in water so they won’t dissolve into condensation inside the pipes. They also must be noncorrosive, so they won’t damage pipelines or appliances, and are formulated to avoid toxic byproducts when the gas is burned.

And they are really pungent. Most odorants contain mercaptans; the main exception, according to Pamela Dalton, a faculty member at the Monell Chemical Senses Center in Philadelphia, is tetrahydrothiophene, or THT, which is widely used by gas distributors in Europe. “It’s still a sulfur compound, but people have described it to me as a little bit more plasticky,” she says. “They say it reminds them of when you open up a fresh can of tennis balls.”

Most gas distributors use blends of at least two odorants, Dalton explains. That’s because local conditions may require special formulas. For example, one common odorant called tert-butyl mercaptan, or TBM, freezes at 34°F, so it has to be blended with other compounds with lower freezing points for use where soil temperatures regularly drop below the mid-30s.

Under federal regulations, odorants must be “readily detectable by a person with a normal sense of smell” at a concentration that’s one-fifth of the lower explosive limit—the lowest concentration in air that would produce a flash of fire in the presence of a spark. Natural gas can catch once it makes up 5% of the air in a space, so gas companies need to ensure a leak can be smelled at 1%. Some states, including New York, Massachusetts, and New Hampshire, require leaks to be detectable at even lower concentrations. According to manufacturers, mercaptans can be detected at concentrations of one part per billion or even less. Residential natural gas typically contains odorants at levels of 0.5 to 10 parts per million to ensure leaks will be quickly noticed.

However, there’s no definition of a “normal” sense of smell, and smelling abilities vary widely from person to person. Aging and illness can degrade people’s faculties. Likewise, there’s no standard for a normal nose. The genes that code for our odor receptors vary by about 30% from one person to another.

Regulations state that odorants “may not be deleterious to persons, materials, or pipe,” but they don’t define allowable concentrations. These compounds are generally considered low risk but can cause eye and skin irritation, respiratory distress, dizziness, and nausea at high concentrations. Workplace safety thresholds for mercaptans, which are also used in chemical manufacturing, oil refining, and even food flavoring, are typically a few parts per million over 8 to 10 hours of exposure.

Pamela Dalton is working with natural gas distributors to reanalyze what concentrations of various odorants people can actually smell. “Thresholds that were measured even 20 years ago have changed,” she says. “Understanding the normal range of sensitivity to these compounds is really important.” If findings show that on average, people can detect odorants at lower levels than gas distributors currently use, the companies may be able to reduce concentrations even further.

The strict requirements for natural gas odorants help explain the dearth of other smell-based alarms. Mining is the only other industry to use odor warning with any frequency. In the case of fire or other threats, mining operations use so-called stench-gas systems to pipe mercaptans through ventilation systems to alert workers. Later, an all-clear odor—typically, wintergreen—signals that the emergency is over.

Mines, like homes and office buildings, are confined systems where a smell can be readily sniffed. Conditions outdoors, however, can change quickly, meaning odors can come and go. Think of the smells you might encounter walking down a busy city street—it could be hard to make potentially critical safety choices based on a passing whiff.

Another factor limiting their wider adoption is people’s tendency to get used to odors over time. That means a smell alert has to be quickly detectable at a very low level. “If it starts to build up slowly, and you’re not paying attention to it,” says Dalton, “your nose will desensitize, and you may not smell it, even though it’s increasing in intensity.” What’s more, if an odor becomes too familiar, people will be less likely to worry when they smell it—what was an alarm becomes part of the background.

Odorants also have an accessibility problem. Many factors can impair our sense of smell, including age. Richard Doty developed the University of Pennsylvania Smell Identification Test, which is widely used for measuring patients’ olfactory ability. Doty has found that about 50% of people aged 65 to 80 have some hyposmia, or smell loss, a rate that rises to 75% for people older than 80. Among that oldest group, as many as 10% to 15% have no sense of smell at all. Smell disorders make it harder for people to detect common hazards, like spoiled food, fire, or gas leaks. They also can reduce patients’ interest in food, which may cause weight loss, malnutrition, and even depression.

Viral infection, head trauma, and pollution exposure are also threats. “Every time we get exposed to a virus, bacteria, diesel exhaust fumes, or nanoparticles, it takes a toll on the epithelium,” says Doty, referring to the tissues lining the nasal cavity. Eventually, that damage affects a person’s ability to smell.

Given how subjective and malleable human olfaction is, some researchers have turned to machines. The idea of building highly sensitive electronic noses that can make fine distinctions among complex smell mixtures was first proposed in 1982 by scientists at the University of Warwick. That proposal has produced a broad stream of research in the decades since. Designs typically use arrays of gas sensors, powered by algorithms that enable them to interpret the data they collect and recognize patterns. Now they are being integrated into robots to be used in search-and-rescue operations, toxic gas leak detection, and food spoilage monitoring. Other researchers are working to make electronic noses that can analyze smells quickly in complex natural settings, such as smoke plumes from wildfires.

Does this mean human olfaction really is diminishing in importance? Far from it.

Smells “give us pleasure, interact with our personalities, and guide our behavior,” writes psychologist Jonas Olofsson, director of the Sensory Cognitive Interaction Laboratory at Stockholm University. “The sense of smell is the most ancient, but perhaps also the most refined of our senses. . . . Most of us have a world of smells to discover.”

Just remember: if you suddenly smell rotten eggs, or a fresh can of tennis balls, leave the room and call your gas company. Captain Mercaptan is counting on you.