CHAPTER 1
Odors in the Mind
It is very obvious that we have very many different kinds of smells, all the way from the odor of violets and roses up to asafetida. But until you can measure their likenesses and differences you can have no science of odor.
—ALEXANDER GRAHAM BELL, 1914
No satisfactory classification of odours can be given.
—Encyclopaedia Britannica, 1911
HOW MANY SMELLS ARE THERE? IT’S AN ODD QUESTION, but give it some thought. Mentally flip through the pages of your personal smell catalog. You find burnt toast, shaving cream, Grandma’s kitchen, and pine trees. There’s the weird glue in the binding of that pocket-size Latin/English dictionary from high school. With a little effort you can come up with a lot of smells, but putting a number to them is difficult. How does one count the odors of a lifetime, much less all the odors in the world?
Some people aren’t daunted by the task: they simply estimate. Better yet, they pass along estimates made by others. Journalists like to say that we can smell thirty thousand different odors. The New Age guru Michael Murphy cites this figure in The Future of the Body (1992): “According to the calculations of one [fragrance] manufacturer, an expert can distinguish more than 30,000 nuances of scent.” Murphy, in turn, got the number from Vitus Dröscher, a German pop science writer (1969): “A perfume manufacturer has worked out that a real expert must distinguish at least thirty thousand nuances of scent.” Dröscher doesn’t provide a source. Perhaps it was in Science Digest (1966): “Industrial chemists have identified some 30,000 different smells.” Unfortunately, the magazine didn’t provide a source either. What this proves, I suppose, is that dubious facts thrived in the media long before the Internet.
One would like to think that smell scientists have a better grasp of the matter, and indeed they prefer to quote a different estimate. When Linda Buck and Richard Axel won the 2004 Nobel Prize for discovering the olfactory receptors, the Nobel Foundation issued a press release. It noted that people recognize and remember “about 10,000 different odours,” a figure the Swedish publicists took from the prize winners themselves. Surely that’s a number we can take to the bank. But the number 10,000 didn’t originate with Buck and Axel: it had been tossed about for years by other scientists. Something about it had always bothered me—why such a nice fat round number? Why was there no date of discovery? And, strangest of all, why did nobody take credit for it?
If you try to track down the mysterious number 10,000 to its original source in the scientific literature, you are in for an adventure; like walking a maze, dead ends abound. For example, I began with a paper in Behavioral Ecology (2001), which I followed to another in Trends in Genetics (1999), which in turn didn’t provide a source. I started again, this time with a prominent smell researcher, the Brown University psychologist Trygg Engen. In 1982 he wrote, “Some have claimed that an untrained person can identify by label at least 2,000 odors and an expert can identify as many as 10,000.” Engen credits this claim to R. H. Wright, Canada’s most famous smell scientist. Wright seemed a likely source, at least until I read what he actually wrote, back in 1964: “[I]t seems likely that the average person would have no trouble in distinguishing between several thousand odours, and an experienced authority in the field has claimed the ability to recognize well over ten thousand. Still another has simply said the number is apparently unlimited.” Wright goes on to say, “It would be an interesting exercise to design an experiment to verify these estimates.” Ooof! So Wright didn’t discover any number at all—he just passed along what he had heard and Professor Engen repeated it. These eminent smell experts remind me of kids at summer camp passing along ghost stories.
I was beginning to think I’d never find the source of the magic number 10,000, when I found it once more in a 1999 food chemistry textbook. From there I followed it to a 1966 paper, and then to a paper published in 1954 by researchers from the Arthur D. Little, Inc., consulting company. At a scientific conference the previous year, the Little folks presented a paper titled “An information theory of olfaction.” Their goal was to place numerical limits on odor perception. They said “There are experts who affirm that it is possible to recognize at least 10,000 odors,” a figure they used in their mathematical subsequent analysis. The name of their expert was buried in a footnote: he was Ernest C. Crocker, a chemical engineer and 1914 MIT graduate who, not coincidently, was also an employee of Arthur D. Little, Inc.
Back in 1927, Crocker and another Little chemist named Lloyd F. Henderson were struggling for an objective way to classify odors. They settled on a method in which an odor was rated by how strongly (on a scale of 0 to 8) it resembled each of four elementary odor sensations. Given the mathematics of their rating system, it was theoretically possible to discriminate 94 or 6,561 different odors. The math is watertight, but the outcome is highly dependent on the initial assumptions. Had Crocker and Henderson used, say, five elementary sensations and a 0-to-10 rating scale, the estimate would have been 115 or 161,051 different odors. (Harvard psychologist Edwin Boring was a fan of the new system, but he believed the rating scale should have fewer steps. He did some calculations and decided that the number of distinguishable smells was somewhere between 2,016 and 4,410.) Discussing this work years later, Ernest Crocker generously rounded up the estimate to 10,000 odors. His colleagues took the number and ran with it.
In the end, it appears that no one has ever attempted to count how many smells there are in the world. Estimates of odor diversity lead either to a dead end or to Ernest C. Crocker. The comfortable, often-cited figure of 10,000 smells is, from a scientific perspective, utterly worthless.
WHY DOES IT MATTER exactly how many smells there are? Suppose we want to build a device that can reproduce every possible odor. (This is a popular fantasy. As a kid you may have scratch-and-sniffed your way through Mickey Mouse and the Marvelous Smell Machine.) A pair of industrial engineers once looked into how many distinct odors it would take to create a lifelike smellscape in virtual reality. They settled on a figure of 400,000. (This number has no more basis in fact than 10,000 or 30,000; its ultimate source is an obscure Japanese technical publication). Four hundred thousand is a staggeringly large number, but it sounds reasonable to engineers who use 16.7 million colors per pixel in visual displays for VR goggles. The trouble is that an engineer’s solution doesn’t always correspond to how the brain solves the problem.
The human eye detects tiny differences in color; across the visible spectrum we are capable of millions of such discriminations. Yet when it comes to naming categories of color, there is nearly universal agreement that only a half-dozen are needed to cover the range of human perception. People in all cultures get by very well with white, black, red, green, yellow, and blue. (Esoteric hues such as ecru and mauve occur mainly in clothing catalogs.) The physical spectrum of visible light is continuous; the stripes of the rainbow are created in our head. We give color names to these few categories.
This simplification of sensory input is a general feature of the brain, a phenomenon psychologists call categorical perception. In hearing, categorical perception helps us carve the continuous dimension of pitch into the individual notes of the musical scale, or the sonic blur of vowel sounds into a distinct a or e. Perhaps we shouldn’t obsess about the number of odors. Our question should be, How many natural odor categories are there, and how do our nose and brain simplify the world?
The Art of the Achievable
I grew up in Davis, California, amid the smells of agriculture. Our house, when we first moved there in 1962, was near vast tomato fields; walking through them brought up the sharp, funky smell of the vines. The approach of a new school year was signaled by the heavy, stewed smell of tomatoes being cooked into ketchup at the Hunt-Wesson plant a mile upwind. My buddies and I played on mountains of newly baled alfalfa, stacked high and smelling grassy-sweet. The playground at the Valley Oak Elementary School offered the hot-metal smell of monkey bars and the dusty, sour resin of tanbark underfoot. The water from the sprinklers in the town park had a musty tinge to it. The office of the Davis Enterprise, where I rolled copies for my afternoon paper route, was saturated with the smell of fresh ink, newsprint, and rubber bands. In grade school my class toured the Spreckles refinery, where truckloads of sugar beets were turned into pure white sugar, a magical transformation dimmed by the suffocating scent of dark molasses that hung over the place.
We moved to Davis because my father joined the faculty of the philosophy department at the university there. Davis was originally the agricultural field station for the main university campus at Berkeley. Set in the hot, flat, and fertile Sacramento Valley, but near the cooler hills of Napa and Sonoma, UC Davis came into its own in the 1960s, when it added a law school and a medical school. At the same time, researchers in the Department of Viticulture and Enology got the ball rolling for what became the California wine-making revolution. They measured microclimates and soil composition, developed new grape varieties, and invented cold fermentation and other wine production techniques. Davis graduates who took courses in wine-tasting and wine-making are now among the world’s leading vintners. As part of this effort, UCD researchers took up the sensory analysis of wine. Their challenge was to apply objective methods to one of the more rarefied arenas of subjective opinion: wine-tasting.
One of this group was Professor Ann Noble, a chemist and sensory specialist. Among her interests was identifying volatile chemical compounds in wine. These substances create the characteristic aroma of grape varieties such as Cabernet Sauvignon or Riesling, and also the off-odors found in poorly made wines. Noble hoped to link aroma chemistry to broader grape-growing and wine-making factors.
Noble’s approach to aroma was practical and effective. Not for her the pretensions of the wine snob, epitomized by Paul Giamatti’s character in the movie Sideways, who sticks his nose in a wineglass and says, “I’m getting strawberries, some citrus…passionfruit, just the faintest soupçon of asparagus, and, like, a nutty Edam cheese.” To better understand California Cabernets, Noble and her colleague Hildegard Heymann selected wines from seven regions around the state (Napa, Sonoma, Alexander Valley, Santa Ynez, etc.). The wines were rated by enology students who came up with their own simple descriptive terms: berry, bell pepper, eucalyptus, and so on. Reference samples were created by doctoring a neutral “base” wine. To represent berry, for example, one-half teaspoon of raspberry jam and one half of a frozen blackberry were added to a half-cup of wine; after soaking for ten minutes, the blackberry was removed. For the soy/prune standard, the basic wine was spiked with a quarter-cup of canned prune juice and seventeen drops of Kikkoman soy sauce.
Armed only with their noses and a nine-point rating scale, the students sniffed and sipped their way to an enormous pile of data. (It filled a metaphorical filing cabinet thirteen descriptors wide, twenty-one Cabernets tall, and thirteen judges deep.) With a computer program, Noble and Heymann extracted a small number of sensory dimensions and placed each wine at a precise location on them. They could now visualize the smell and taste relationships between the samples. Their conclusions: “Younger vines and/or vines from cooler areas tend to produce more intensely vegetative wines. Conversely, wines from older vines and/or warmer areas tend to have higher ratings for berry aroma, fruit flavor by mouth, and vanilla aroma.” By quantifying wine aromas with a system of practical description, they discovered the vineyard conditions that produce them. Alexander Graham Bell would have been pleased.
WHAT EARNED Ann Noble and her colleagues a place in the annals of smell classification was their Wine Aroma Wheel, published in 1984. The wheel was a visually pleasing presentation of a standardized wine aroma vocabulary. With twelve categories and ninety-four descriptive terms, it covers the aroma of any wine, regardless of grape or geographic origin. What makes the wheel just as useful to novices as to connoisseurs are the do-it-yourself reference standards. Simple kitchen chemistry lets anyone create and experience the standards for himself. Noble is critical of commercial wine-tasting kits; she believes that vials of flavor essence are chemically unstable and tend to degrade quickly. This spurred her to create reference standards that can be “prepared using foodstuffs available throughout the world during most seasons.”
The Wine Aroma Wheel resembles a dartboard: three concentric circles divided pizzalike into a dozen slices of varying width. On the innermost circle, the pointed end of each wedge is an aroma category, such as fruity. In the middle circle, the wedge may be split into subcategories such as citrus, berry, or tree fruit. On the outer circle are specific materials, examples of each aroma subcategory. Thus you can follow the fruity wedge through the berry subslice to the outer circle; there you’ll find blackberry, raspberry, strawberry, and blackcurrant. The beauty of Noble’s wheel is that it links sensory concepts to actual everyday stuff—it connects Riesling to raspberries. Wheel in hand, it is possible to sniff your way to sensory enlightenment. This commonsense approach lets anyone grasp the esoteric innerwedge category microbiological and its arcane subdivision lactic. It’s only baffling until you sniff the examples: yogurt and sauerkraut. Then it clicks. The wheel even demystifies the wine-snob term wet dog: it’s an example of sulfur aromas in the chemical category (along with skunk, cabbage, and burnt match).
There is no place on the wheel for the wine critic’s gaseous adjectives. You will find “orange blossom” and “black olive” and the less flattering “soapy” and “cooked cabbage.” But you will not find “an impertinent little Pinot Noir” or a “flabby, overripe Cabernet Franc,” à la Miles Raymond. These free-form prose poems say more about a wine lover’s pretensions than about the character of the wine. To use the wheel, all you need is a glass and grocery store.
A PRACTICAL SMELL classification for beer was created in the 1970s by a Danish flavor chemist named Morten Meilgaard. His Beer Flavor Wheel has now been adopted worldwide. It uses fourteen categories and forty-four sensory terms to describe the smell and taste of any style of beer—lager, ale, or stout. Most of the descriptors deal with aroma; others involve taste (bitter for hops; sweet for malt) and sensory factors like carbonation. Meilgaard’s system includes reference standards, but unlike Noble’s wine wheel (which was inspired by it), one needs access to pure chemicals to create them. For example, to mimic the “papery” aroma of oxidized beer, one doses a pitcher of beer with trans-2-nonenal.
A brewer’s best friend is his nose. Desirable aromas tell him when the product is on target. By identifying off-smells in the product, a brewer can correct problems in the brewing process. For example, the smell of wet newspaper indicates that a beer has oxidized. Sunlight-damaged beer has a skunky smell. (Many years ago, Corona beer from Mexico was poorly made and oxidized easily. The acid in a slice of lime was an effective way to chemically neutralize the off-odor. Today Corona is made as well as any beer in the world, but the lime tradition lives on.)
Meilgaard’s beer system is less satisfying for the lay drinker than the Wine Aroma Wheel. Because the reference standards are made with single chemicals, they can be prepared with great precision. However, they don’t reproduce complex aromas like raspberry or asparagus, and they are not cheap and easy for the amateur beer enthusiast to make at home. An added frustration is that the descriptive terms for beer are confusing. In the “sulfury” category, for example, are “sulfury,” “sulfitic,” and “sulfidic,” terms only a chemist could love.
THE APPEAL OF aroma wheels is that they organize product-specific smells into a few, easily recognized categories. As a result, food lobbies around the world have come up with their own versions. There is a chocolate aroma wheel from Switzerland and a Flavour Wheel for Maple Products, courtesy of Canada. There is a pan-European wheel for Unifloral Honey, and another for cheese (although with seventy-five different aromas it doesn’t really simplify things for cheese fans). There is a South African brandy wheel, and the Berkeley-based perfumer Mandy Aftel has created a Natural Perfume Wheel. Recently, some guys in the Philadelphia Water Department came up with a wheel for identifying the odors found in sewage. (Anyone who’s lingered on the banks of the Schuylkill River knows that wastewater offers a particularly rich olfactory experience.) The world has gone wheel crazy, and we can expect to see more of them in the future.
The Perfumer’s Problem
Odor space is an imaginary mathematical realm containing all possible odors. The aromas of wine and beer occupy only a small fraction of odor space—not nearly the full range of smells detectable by the human nose. Can smell classification work on a larger scale? Perfumes and colognes take up a bigger chunk of odor space: There are at least 1,000 currently on the market, with new ones added at a rate of about two hundred per year. Each has anywhere from 50 to 250 ingredients. If anyone has a lot of smells to keep track of, it is the perfumers who create them.
At its core, the practice of perfumery hasn’t changed much since it came to full flower in Renaissance Italy. In those days there were no more than 200 commonly available ingredients, all derived from natural sources, either botanical (essential oils, gums, spices, and barks) or animal (musk and civet). By the late nineteenth century, discoveries in organic and synthetic chemistry created a host of new materials. Some were novel, man-made molecules; others were pure chemicals isolated from the complex mixtures found in nature. The result is that the modern perfumer’s palette is far larger than his predecessor’s. Learning these materials is a correspondingly bigger task. How does a perfumer keep it all straight?
The professional perfumers Robert Calkin and Stephan Jellinek explain: “The novice perfumer may well feel daunted by the hundreds of bottles containing strange and often unpleasant smelling materials that line the laboratory shelves. But for the talented student the task of learning to identify them is in fact less difficult than it may seem at first.” The trick, according to these experts, lies in honing specific cognitive skills, namely learning new mental categories and how to fit new smells into them. To become a perfumer you don’t learn to smell like one—you learn to think like one.
The first step in training is to learn the smell of the available ingredients. The leading teaching technique—the Givaudan method, created by the French perfumer Jean Carles—introduces students to the major ingredients using a matrix approach. Imagine a grid of rows and columns. Each row is a fragrance family: citrus, woody, spicy, and so on. Each column is a training session. In the first session, students smell column-wise one material from each family: lemon oil, sandalwood oil, and clove bud oil, for example. In the second session, the students smell new examples: bergamot oil, cedarwood oil, and cinnamon bark oil. This process continues for about nine lessons, by which time the students are familiar with the olfactory differences between families. Now comes the hard part—learning the “contrasts” within a family. Each subsequent session traverses one row of the matrix. In the citrus lesson, for example, students smell lemon, bergamot, tangerine, mandarin orange, blood orange, grapefruit, and lime. The goal, according to master perfumer and teacher René Morgenthaler, is for the student to create a personal impression of each ingredient. These individualized mental hooks are the key to remembering the fine discriminations needed to do perfumery. The graduate of nasal boot camp must recognize more than 100 natural materials and around 150 synthetics. The professional perfumer eventually becomes familiar with every material in his company’s library—anywhere from 500 to 2,000 items—and is able to recognize every grade of each.
With the basic raw materials in mind, a trainee next learns to think like a perfumer. When a professional analyzes a fragrance or creates a new one, he does not think in terms of individual ingredients; he thinks of typical combinations called accords. An accord is a mixture of raw materials (rarely more than fifteen) that go together particularly well. Accords are the building blocks of perfumery. By combining several of them, the perfumer creates an initial sketch of the perfume, sometimes called a skeleton. In a way, creating a perfume is like writing software: a programmer starts with building-block software modules that already contain many lines of code. A computer program is built with many modules, just as a fragrance is assembled from accords. The analogy goes further—software is tested with iterative debugging; perfume is tested with repeated sniffing and tweaking of the formula.
An art form as subjective and personal as perfumery might be expected to resist computerization. In fact, the opposite is true. The practice of perfumery quickly adapted to the digital world in terms of tracking materials and recording formulas. At a more fundamental level, the perfumer and the software programmer share a similar mind-set that involves the logic of subprograms and modules. Some of the memory burden of those thousands of ingredients is relieved by computer technology. Perfumers browse the company’s entire inventory of materials on-screen. They assemble a formula with a series of mouse clicks. They save everything: formulas, failed trials, and favorite accords. Software is an active partner in the creative process. It warns the user when two chemically incompatible materials have been selected, thereby avoiding a formula that discolors when exposed to sunlight. Most important, it continuously tallies the cost of the formula and displays it on-screen as dollars per pound of fragrance oil. No matter how great the creative latitude on a given project, a perfumer always works to a dollar limit.
Once a novice starts to think like a perfumer, he begins to develop a new way of smelling. Individual ingredients recede and whole fragrances emerge: he learns to smell the forest before the trees. Given a new men’s cologne, he quickly recognizes it as, say, a Fougere type. Next he sniffs for the individual notes that define the Fougere pattern: lavender, patchouli, oakmoss, and coumarin. After confirming these, he smells further, looking for a new twist or nuance that sets this formula apart from all the other Fougeres in the world.
Perfumers reduce the complexity of their world to a small, manageable number of fragrance families. They use well-known accords to simplify the process of scent creation. The perfumer’s job is more about pattern recognition than about raw memorization; his mental map is uncluttered by free-floating details. Like most highly creative people, perfumers tend to be a little crazy; but they are not driven crazy from remembering thousands of smells.
The Shopper’s Problem
Hundreds of perfumes are available for sniffing in department stores and boutiques. They range in style from restrained elegance to loud assertion, from distinctive originals to blatant knock-offs. How does one shop for scent amid this sensory overload? Lacking the perfumer’s trained thought processes, the average person is completely at sea.
Fragrance houses—the companies that employ perfumers and create the juice for the Calvin Kleins and Cotys of the world—find it useful to organize perfumes by smell. The Haarmann & Reimer company published a fragrance genealogy that traces every style of perfume from its first appearance to the present day. It’s a nasal Book of Genesis: In the beginning was Jicky (Guerlain, 1898), and Jicky begat Emeraude (Coty, 1921), and Emeraude begat Shalimar (Guerlain, 1925), and so on through Obsession (Calvin Klein, 1985) and those that followed it. (Although some ancestral scents are well-known classics, it is sobering to see all the brand names that meant so much in their time but so little today: Moon Drops (Revlon, 1970), Touché (Jovan, 1980), or Aspen (Quintessence, 1990). Genealogies provide a sense of history, but they don’t help one to shop in the here and now.
Another style of perfume guide lists each brand by fragrance family: florals, aldehydics, chypres, and so on. This isn’t much help if you don’t know what a chypre smells like. (The term covers a range of styles having in common a warm, woody character with an animal-like undertone.) If you like Estée Lauder’s Pleasures, you can look up a dozen similar scents. What you won’t find is a measure of how similar they smell. Nor will you find the exact ways they differ from Pleasures—are they stronger, spicier, greener, muskier?
Most people don’t consult a reference book before shopping. They simply head for the department store. But once inside, things don’t get easier. Each fragrance brand has its own counter, attended by its own salesperson who will show you only the perfumes she is paid to show. If your ideal scent is one counter away, it might as well be in a different universe. Sephora stores broke with retail tradition by introducing the “open sell.” Brands are arranged alphabetically on the shelf, from Alan Cummings to Yves Saint Laurent. With no vested interest in any one brand, the Sephora staff is just as happy to sell you Alan as Yves. To introduce sensory logic to their store designs, the company has tried arranging perfumes by fragrance family: orientals here, florals over there. This may be the start of a badly needed rethinking of the retail experience.
Charts and guides, even those based on expert opinion, are still arbitrary views of odor space. They present the world according to one fragrance house, or more likely just its chief perfumer. No single classification has emerged thus far as the industry standard. If one did, it still wouldn’t help the average shopper, because perfumers don’t think like the rest of us. The professional detects rose de mai Bulgarian where the consumer smells flowers. The professional finds clear-cut differences among perfumes that strike most people as indistinguishably fruity-floral. What the average person needs is a map on which brands are arranged by how they smell to other average people.
PERFUME MAKERS speak to the consumer with two voices: Ingredient Voice and Imagery Voice. Here is a classic example of the Ingredient Voice, from a description of Estée Lauder’s best-seller Beautiful (1985):
Vibrantly feminine floralcy of rose, lily, tuberose, marigold, muguet, jasmine, ylang, cassis and carnation accented with fresh mandarin and bright fruity notes. Warm background accord of orris, sandalwood, vetiver, moss and amber.
Ingredient Voice assumes perfumer-level familiarity with more than a dozen raw materials, when in fact few civilians have ever smelled orris root or vetiver. Reciting a list of ingredients gives an illusion of precision. Even perfumers don’t think of Beautiful as a list of ingredients; they might think of it as a big, complex floral type with an ambery warmth. Ingredient Voice doesn’t help the casual shopper.
In contrast, Imagery Voice is all about atmospherics. The drama of seduction, passion, and mystery makes Imagery Voice the natural language of brand marketers and ad agencies. Listen to an actual vice president of marketing discuss a new men’s cologne with a cosmetics industry trade magazine: “It’s intended to target a young, stylish, hip, contemporary kind of guy.” So far, so good. Aging, badly dressed nerds aren’t known for buying a lot of cologne.
“The positioning of [the new brand] is really based all about capturing the pulse and energy of the city.” Reasonable enough—discretionary consumer dollars don’t chase listless, slow-moving, rural scents. But what does the new scent smell like? Hearken to the Imagery Voice:
The fragrance notes themselves are city inspired, in that the top notes we describe as being powered by “living liquid air.” It’s fused into a matrix of metal aldehydes and it captures the feeling of shiny steel and glass in a modern urban environment. It is very fresh and almost metallic on top; then it dries down to warmer, more sensual suede and woody notes on the bottom.
That’s an impressive prose poem. Buried in it are two actual smells: suede and wood. Suede is fairly specific—one imagines the smell of a supple jacket or new pair of shoes. Woody, on the other hand, covers a lot of territory: pine, oak, cedarwood, redwood, cypress, and don’t forget sandalwood. If Mr. Young Stylish and Hip wants to know what this new cologne smells like, he’s just going to have to smell it for himself.
Imagery Voice combines ordinary adjectives (fresh, woody) with technical terms (aldehydes) and envelopes them in emotional verbiage (“the feeling of shiny steel”). The result is the marketing equivalent of a Jell-O mold at a church dinner.
NOTICEABLY ABSENT from the world of perfume is the identifiable voice of the independent critic. There is no Roger Ebert of scent. A scientific polymath and self-appointed expert named Luca Turin once tried his hand as a freelance perfume reviewer. Turin is serious about the aesthetics of fragrance and not a shill for any manufacturer. But his capsule reviews tend to be highly stylized. Here’s a whiff: “Après l’Ondée evolves only slightly with time: its central white note, caressing and slightly venomous, like the odor of a peach stone, imposes itself immediately and retains its mystery forever.” Turin makes Après l’Ondée sound both impossibly abstract and off-puttingly tactile. Meanwhile, a reader still doesn’t know what it smells like.
Perfume wearers need a style of commentary that blends the aesthetic and the technical, like the road tests in Car and Driver that talk about sporty handling and trunk space in the same paragraph. In 2006 the New York Times tapped Chandler Burr as its first-ever perfume critic. Burr rates perfumes with a conventional five-star scale and a writing style that tilts heavily toward the aesthetic: “This is the scent of the darkness that inhabits a Rubens, a warm, rich, purple blackness; Pomegranate Noir is like a box of truffles with the lid on, sweet bits of darkness, waiting.” (OK, but what about the horsepower and mileage?)
Practical-minded perfume fans might prefer “Andrew,” who pens cheeky analysis for the English newspaper Metro. Here’s his take on Live Luxe by Jennifer Lopez: “It’d take a very brave/mad woman to wear this one. Ridiculously sweet and fruity, this is the fragrance equivalent of going out dressed as Carmen Miranda with a fruit cocktail poured down your cleavage. Invigorating but not for use in an enclosed space.” Andrew recommends it “[f]or ladies who like to make an impression.”
How come we have Cigar Aficionado and Wine Spectator, but no Perfume Enthusiast? This is a magazine publishing niche waiting to be filled. In the meantime, perfume bloggers are popping up all over the Internet: IndiePerfumes, Anya’s Garden of Natural Perfumery, SmellyBlog, Scentzilla, to name just a few. As elsewhere in the blogosphere, this evolving community is a mixture of the personal and the professional, the serious and the whimsical. But the passion for fragrance is always there. These writers are pioneering new ways of describing scent. I think their efforts may produce a vibrant, robust, and very useful way of organizing the world of perfume.
The Big Enchilada
Perfumes, flowers, and wine occupy the sunny heights of the smellscape. Beyond lies the Dark Side, a swampland reeking of burnt rubber, rotten eggs, and the silent but deadly guy on the No. 33 bus. Few people aspire to study stench—there are no maestros of malodor. And yet, if we are truly to understand the sense of smell, we must account for the whole of it: the good, the bad, and the ugly. Where is the Universal Classification of Smell?
According to conventional wisdom, all major smell classifications can be traced back to the eighteenth-century Swedish naturalist Carl von Linné (1707–1778), known as Linnaeus. Linnaeus was the Big Daddy of scientific classification. In fact, he was a little obsessed with the topic: he classified plants and animals, rocks and sea creatures, and even his fellow scientists. Far from being a muddy-boot field biologist, Linnaeus was a bookish desk-jockey more concerned with defining the single “type” of a species than with the extent of natural variation. For this reason, some historians view him as a rigid essentialist who held back progress in the life sciences for decades. Still, his decision to assign a two-part Latin name to every species—something he regarded as a minor innovation—was a stroke of genius, and it became the basis of all modern taxonomy.
Linnaeus is widely credited among psychologists with inventing the first scientific classification of smells. Very few of them, however, seem to have read the actual treatise, published in 1752. Its Latin title, Odores medicamentorum, translates as “The smells of medicines,” and this is the first big hint that Linnaeus’s primary interest was not smells, but the medicinal properties of plants. He believed he could predict the therapeutic effect of a plant from its odor. To his way of thinking, nonsmelly plants were medically worthless, while strong-smelling ones had great pharmaceutical potency. Similarly, he believed sweet-smelling plants were wholesome, nauseous ones were poisonous, spicy ones were stimulating, and “noisome” ones were “stupefying.” These effects were due to plant smells acting directly on human nerves. You can be forgiven if the views of Sweden’s greatest scientist sound to you like those of a New Age aromatherapist in contemporary Santa Monica.
In grouping medically useful plants by odor, Linnaeus came up with seven classes that translate as fragrant, spicy, musky, garlicky, goaty, foul, and nauseating. His only concern was using smell to classify natural medicines; he did not intend to create a universal classification of all smells. In fact, he had little interest in smells as smells. (This explains the absence of such obvious odor categories as floral, fruity, woody, and leafy green.) Despite his focus on medical properties and his neglect of sensory qualities, European scientists viewed Linnaeus as the first scientific classifier of smells, and the results were a disaster—it sent smell researchers on a wild-goose chase that lasted for two centuries.
The next scientific smell classifier emerged toward the end of the nineteenth century. The Dutch physiologist Hendrik Zwaardemaker (1857–1930) was, by his own account, not particularly interested in smells. His lack of feeling for the topic shows in his work, where his main contribution was to add two new classes (“ethereal” and “empyreumatic”) to those of Linnaeus and to create subdivisions within each class. The new version was complicated and made little sense as a comprehensive classification. (He was, after all, cramming every smell in the world into categories meant to organize only smelly medicinal plants.) Zwaardemaker labored to explain his system, but his tedious cross-referencing of previous classifications has all the prose sparkle of the IRS tax code. Like the system it expanded on, Zwaardemaker’s classification was based entirely on one man’s opinion, rather than on experimental data.
The German physiologist Hans Henning (1885–1946) relentlessly attacked the inconsistencies and absurdities in Zwaardemaker’s classification. He took aim at Zwaardemaker’s preference for lifting odor descriptions from novels and literary works rather than from the direct experience of his own nose. Henning insisted that sensory experience was superior to empty intellectualizing; his motto was “just smell it.” His own classification, proposed in 1916, had two very important selling points: it was based on empirical data, and it came with a ready-made visual representation, the “odor prism.” The image was a compelling one, orderly and neatly geometric. The six corners of the prism were each assigned a specific odor quality. Henning claimed that any odor could be located on the surface of the prism; its distance from any corner indicated the relative contribution of that odor quality.
Unfortunately, Henning overplayed his hand. The clean geometry of the odor prism proved irresistible to the scientific psychologists in America, who tested its feasibility in laboratories at Harvard, Clark, and Vassar. Initially enthusiastic, the Americans soon found his theory to be cumbersome and too vague to yield testable predictions. In their hands, it produced inconclusive results. Henning’s initial theory was based on work with only a few experimental subjects; it now became clear that those subjects were extremely, if not suspiciously, consistent in their responses. (Wide person-to-person variability is a hallmark of odor perception; it’s unlikely that randomly selected sniffers would agree as precisely as Henning’s trio did.) In retrospect, there was always something too neat about Henning’s idealized prism: its geometric elegance is undeniably appealing, but few areas of human experience are less linear than smell.
The dismantling of the odor prism by American psychologists ended the European tradition of armchair smell taxonomy. The search for a Universal Classification of Smell shifted entirely from philosophical reasoning to experimental research, and with it momentum crossed the Atlantic for good. Although as outmoded as the buggy whip, the odor prism persists in contemporary encyclopedias and textbooks, a testament to its iconic power.
IT WAS FRUSTRATION with Henning’s prism that led the Americans Ernest Crocker and Lloyd Henderson—of the “10,000 odors” estimate—to invent a new system of smell classification. They began by selecting four “elementary odor sensations”: fragrant, acid, burnt, and caprylic. Then they assembled a set of odors to serve as reference standards, by means of which any smell could be rated on a scale of 0 to 8 for each of the elementary sensations. Rose, for example, was rated 6 on fragrant, 4 on acid, 2 on burnt, and 3 on caprylic. Those four numbers (6423) became, presto change-o, a digital identifier for that particular smell. In the same way, vinegar was 3803 and freshly roasted coffee was 7683. A numerical specification of sensory quality is not that outlandish; the Pantone color standards, for example, use numbered samples to let graphic designers and printers communicate accurately.
The Crocker-Henderson system had wide appeal because it was based on empirical data and an open set of standards: anyone could play. Following its publication in 1927, the system was quickly commercialized; the complete set of reference odors could be ordered from Cargille Scientific, Inc., in New York City. It was soon being used by distillers, soap companies, the U.S. Army, and even the Department of Agriculture. Sensory psychologists initially gave the system positive reviews, but in 1949 researchers at Bucknell University dealt it a stunning blow. They found that untrained people couldn’t sort the thirty-two reference odors into anything resembling the four elementary sensations postulated by Crocker and Henderson. Further, people were unable to arrange the eight odors within an elementary group in order of intensity. Because the Crocker-Henderson system was premised on elementary odors and intensity-graded smells within them, the new findings effectively undermined its logic. User enthusiasm vanished and the system eventually faded away.
ANOTHER BURST OF innovation in odor classification took place in the 1950s and 1960s when chemist John Amoore observed that people who were odor-blind to the stinky-feet smell of isovaleric acid were relatively insensitive to similar smells. He proposed that “sweaty” was a primary odor in the same way that red is a primary color. Amoore sought out molecules with similar shapes and smells that he thought might be the basis of other primary odors. (He eventually proposed seven of them: camphoraceous, musky, floral, pepperminty, ethereal, pungent, and putrid.) While he did succeed in finding other instances of selective odor blindness, Amoore’s notion of primary odors did not hold up under rigorous sensory testing. In the end, the structural features of a molecule are not a reliable guide to the psychological realities of odor categories.
The latest attempts at odor classification use a technique called semantic profiling, an approach pioneered by the fragrance chemist Andrew Dravnieks in the 1960s, and still used today. Researchers hand people a long list of smell descriptors and have them check off as many as apply to a given odor sample. The hope is that with enough descriptors, smells, and statistical analysis, a pattern will emerge. And indeed patterns do appear—it is possible to point to groups of odors that share similar descriptions. The trouble is, this leaves us back where we started from—odors are described similarly because they smell similar. What we really want to know is, Why do they smell similar? For now, scientists are stumped—the molecular structure of odors isn’t the answer, nor can we conjure categories from lists of adjectives. As a result, researchers today are reluctant to propose anything like the grand classifications of the past.
IF HISTORY IS littered with the wrecks of Universal Classifications of Smell, we can still learn something from surveying the ruins. What they have in common is a surprisingly limited number of elementary categories: either 4, 6, 7, or 9, depending on who you like. The mind-boggling variety of smells in the world is reducible to a manageable handful of nameable odor classes, just as the brain carves the range of visible light into a handful of focal colors. Suppose one adopted the standard perfumery categories as an approximation of the pleasant sectors of smell space; this amounts to one or two dozen classes (woody, floral, fruity, citrus, etc.). What more would one need to encompass the stinks and stenches of the world? The fecal category would cover a lot of territory—from benign horse manure to the intolerable air in a rock concert privy. A category for urinous could include the sour smells in a nursing home and the heavier fourth-quarter reek of urinals at an NFL game. We’d have to add a class for retch-inducing smells—vomit and really stinky feet—and another for fishiness in all its gradations. Skunk, sulfur, and burning rubber could constitute yet another class. Finally, the putrid stench of rotting meat probably deserves its own banner. These six classes would capture most of the bad smells abroad in the world. Which is more amazing—the huge number of possible odors, or the tiny number of odor types?
CAN SUCH A stripped-down system of classification handle the olfactory complexities of the real world? It turns out that the human brain already does a pretty good job of reducing that complexity. The Australian psychologist David Laing was the first to tackle the relevant question: How many smells can we pick out of a complex mixture by nose alone? He began with a set of distinctive odors such as spearmint, almond, and clove, each easily identifiable on its own. He created mixtures—beginning with combinations of two odors at a time—and asked people to identify as many components as they could. The more odors he added to the mixture, the more difficult it became to identify even a single ingredient within it. The degree of difficulty was surprising. For example, in a mixture of three or more odors, fewer than 15 percent of people could identify even one component. Laing made the test easier: he gave people a target odor and asked them whether they could smell it in the mixture. Even then, they could rarely find the target in a mix of more than three odors. Could the problem be lack of skill? Laing tested perfumers and flavorists. The professionals were better than amateurs at identifying two and three items in a mixture, but even with their training and experience, they failed to pick more than three odors from the mix. Laing reasoned that mixtures of simple, single-chemical smells are somehow unnatural and hard to pick apart. So he repeated the experiments using as mixture components such complex odors as cheese and chocolate. The results were the same: no one could bust the four-odor limit. Were the individual smells not distinctive enough? Some odors, such as orange, almond, and cinnamon, blend together easily; perhaps those that blend poorly, such as mushroom, cut grass, and mandarin, are easier to pick out of a mixture. Laing found this was true to a point, yet the four-item barrier held firm.
Why are we so feeble at smelling our way through a bouquet? Our ability to gather olfactory information is formidable: the human nose detects single smells at extraordinarily low concentrations. We do a better job of collecting smells than we do of tracking them in a complex mixture. The Laing Limit suggests that the problem is not in the nose but in the brain. We have limited ability to think about smells analytically.
In the end, the question “How many smells are there?” may not be as relevant as “How many odor categories do we need to make sense of the world?” The answer to that question will reveal much more about how the brain handles the information that the nose provides.