Những người đã cố xếp mùi vào ô

Colour has wavelength. Sound has frequency. Smell has no natural scale of that kind, so for almost three centuries each period has built its own frame to sort smells into. This essay walks through those frames, by checked year of publication.

An eighteenth century list of smells

In 1752, at Uppsala, Carl Linnaeus presided over a medical dissertation called Odores medicamentorum, defended by Andreas Wåhlin. It was reprinted in 1756 in volume three of Amoenitates Academicae, which is why many sources give 1756. Its purpose was not perfume. Linnaeus wanted to infer the use of a remedy from its smell, in line with the medicine of the time.

The dissertation divides smells into seven classes: aromatic, fragrant, ambrosial, alliaceous, hircine, foul and nauseous. The first three are pleasant, the last four are not. It is among the earliest systematic scent classifications still cited.

Later systems fall into two lines. The scientific line looks for the structure of the sense of smell itself. The perfumery line arranges materials and perfumes for work and for sale. The two lines cross many times, but they are not asking the same question.

Rimmel and the first scent families

In 1865 Chapman & Hall in London printed The Book of Perfumes by Eugène Rimmel, born in France and established as a perfumer in London. Most of the book is a history of aromatics across cultures. Rimmel's classification takes only a few pages of the first chapter, yet it is the part cited most often later.

Rimmel borrowed from painting. There are primary colours from which every shade is mixed; he argued that there are also primary odours with a perfect type, and that other odours are tied to them more or less closely. The table has 18 classes, each with three columns: the class name, the type odour, and the odours placed in the same class. Rose goes with geranium; vanilla with balsam of Peru, Tolu, benzoin and tonka; anise with star anise, caraway, dill and coriander.

The table covers pleasant odours only. Rimmel recorded two limits himself: some odours fit no class, and his example is wintergreen; and the table does not cover the scents that arise when several classes are blended. The almond class even holds mirbane, that is nitrobenzene, a synthetic that Rimmel himself called the most widely used artificial material in the trade of his day.

Just before the table, Rimmel faults Linnaeus for putting carnation beside laurel leaf and saffron beside jasmine. Linnaeus grouped odours to infer the use of a remedy; Rimmel grouped them to work with them. A compounder needs to know which material stands nearest when one is missing. The column of related odours reads like a list of substitutes or companions.

In the last chapter Rimmel uses an entirely different division: by source. The animal series, the floral, the herbal, the ligneous, the roots, the seeds, the balsams and gums, the artificial. The floral series then held only eight usable flowers: jasmine, rose, orange, tuberose, cassie, violet, jonquil and narcissus. Division by source later became the usual order of raw material catalogues; division by odour led to the scent families.

Piesse and the gamut of odours

Septimus Piesse (1820 to 1882) was a London chemist and perfumer and a co-founder of Piesse & Lubin. His The Art of Perfumery was first printed in London in 1855; the 1857 Philadelphia edition is the one on Project Gutenberg, which is why many sources give 1857.

In the 1855 edition the idea takes only a few sentences. Odours act on the olfactory nerve in definite degrees, like sounds; there is an octave of odours in which certain odours coincide like the keys of an instrument. Almond, heliotrope, vanilla and orange blossom give one impression; citron, lemon, orange peel and verbena make a higher octave. Right after, in a recipe for sweet pea, Piesse explains why he fixes it with vanilla rather than musk: vanilla strikes the same key as orange blossom, so as the ingredients evaporate the scent on the handkerchief does not turn into a different idea.

The third London edition of 1862 expands this into a chapter. Piesse's gamut is printed on two clefs: the treble from F (civet) down to D (violet), the bass from C (rose) down to C (patchouli). In 1887 his son Charles Piesse named the gamut the odophone.

The gamut does not hold as science. Sound has a measurable quantity, frequency, and an octave is a doubling of it. Smell has no matching quantity. An odour's place on Piesse's gamut rests only on his own perception, and he gives no way for anyone else to check it.

What remains of Piesse is vocabulary. The 1862 chapter already has the whole set of words: key, note, chord, harmony, discord, treble clef, bass clef. Perfumers today still speak of top, heart and base notes, and call a group of materials forming a new impression an accord.

Zwaardemaker and the laboratory

In 1895 W. Engelmann of Leipzig published Die Physiologie des Geruchs by Hendrik Zwaardemaker (1857 to 1930), a Dutch physician and physiologist, then an army medical officer and lecturer in Utrecht. The book runs to more than three hundred pages, mostly on measurement: thresholds, intensity, fatigue of the nose, compensation between two odours.

Zwaardemaker's nine classes keep Linnaeus's frame of seven and add two: the ethereal class, for fruity esters, and the empyreumatic class, for the smell of smoke, tar and benzene. He explained that the old seven classes had been drawn up for the remedies of Linnaeus's day; modern chemistry needed room for new odours.

What is new is the column of examples. Rimmel took one material as the type of each class; Zwaardemaker, where he could, took a compound of known structure. The aromatic class is split into five subclasses, each with a type compound: borneol for the camphor group, eugenol for the spicy group, anethole, menthol and thymol for the anise and lavender group, geraniol and citral for the lemon and rose group, benzaldehyde for the almond group. He cites Rimmel many times; a perfumer's table entered a physiology textbook.

Zwaardemaker also recorded where things did not fit. Some oils contain esters yet had been placed in another class by smell. His remark: similar odours usually go with analogous composition, but substances of analogous composition do not necessarily smell alike. That question is still open.

The instrument that made his name predates the book by seven years: the olfactometer, first described in 1888. A curved glass tube is placed in a nostril; over it slides a cylinder 10 cm long, made of or coated with the odorous material. The further the cylinder is drawn out, the more odorous surface meets the inhaled air. For the first time, the strength of an odour stimulus could be written down as a number that could be repeated.

The modern fragrance wheel

The twentieth century tried more geometry. In 1916 the German psychologist Hans Henning placed six primary odours at the six corners of a triangular prism: flowery, fruity, spicy, resinous, burnt, putrid. In 1927 Crocker and Henderson described every odour on four scales from 0 to 8, giving a four digit code; vanillin is 7122. In 1951 the perfumer Paul Jellinek arranged odours by perceived effect on two opposed axes.

The perfumery line reached a more stable form. The Technical Commission of the Société Française des Parfumeurs drew up its first classification in 1983, published in 1984, for women's fragrances only, in five families: floral, chypre, fougère, amber, leather. The 1990 edition added citrus and woody, making seven scent families. According to the SFP, the current version has seven families and 47 subfamilies across women's, men's and shared fragrances.

In the same year, 1984, Michael Edwards, then working in Sydney, self-published The Fragrance Manual, a guide to help retailers suggest perfumes. The 1992 edition was the first to include a fragrance wheel: families arranged in a circle, each passing gradually into its neighbour. In 2000 the guide was renamed Fragrances of the World, and the wheel has been revised several times since. Many pages date the wheel to 1983; checked sources show the manual in 1984 and the wheel first printed in 1992.

The wheel suits retail because it shows a path, from a liked scent to its neighbours. It inherits Henning's idea of continuous transition, but drops the ambition to describe the nature of smell. It is a map for finding one's way, not a law.

What every system leaves out

In 1991 Linda Buck and Richard Axel reported in the journal Cell a large gene family encoding odorant receptors. Each receptor responds to a group of molecules; each molecule activates a combination of receptors. Smell is coded by combination, not by a few primaries. The two shared the 2004 Nobel Prize in Physiology or Medicine.

This result explains why no table of primary odours maps neatly onto biology. Seven classes, nine classes, six corners, four digits, Amoore's seven primaries: each number is a way of cutting a space opened by some 400 receptors into a few boxes of manageable size. A 2025 review in Chemical Senses concludes that no universal relation between odour quality and physical or chemical parameters has been found.

Each system also carries the mark of its period's tools. Linnaeus had a medicine cabinet. Rimmel had the shelves of a London shop. Piesse had the musical staff. Zwaardemaker had the olfactometer and the structural formula. The SFP and Edwards had a perfume market with thousands of names. Whatever lay outside the reach of the tool stayed out of the table: unpleasant odours for Rimmel, blended odours for most tables, the difference between two noses for nearly all of them.

The Lê Mai library arranges scent families on the SFP frame, which places it in the perfumery line, not the scientific one. The choice is deliberate: a shared frame lets perfumers and readers speak the same language. It makes no claim to have found the structure of smell. The essay The Families of Scent goes into each family in that frame.

Each frame holds one part.

The rest stays outside, in the air.

The name comes after the smell, and is always smaller than it.