Headspace
To smell a flower, perfumery had to kill it. Then somebody put a glass bell over a living blossom and took only the air.
Every classical method of capturing a scent destroys its source. Distillation boils it, extraction dissolves it, enfleurage exhausts it onto fat. All of them return a material that is genuinely beautiful and is not what the living plant was emitting. From the late 1970s a different approach arrived: enclose the living flower in glass, draw its air through an adsorbent, and analyse what was actually in the air. Headspace made it possible to record a smell instead of harvesting one, gave perfumery flowers it had never been able to obtain at all, and quietly changed what the word natural can be made to mean.
A jasmine flower on the vine at four in the morning is not the same object as jasmine absolute in a bottle, and for most of the history of perfumery nobody could prove it. The flower was picked, and from the moment it was picked it began to become something else. Whatever it had been doing while alive was gone before the material reached anybody's nose.
This was accepted as the cost of doing business. Perfumery is a subtractive art: to have the smell you must end the thing that was making it. Enfleurage is the most explicit about this, laying blossoms on fat until they are spent and replacing them with fresh ones, over and over, for weeks.
Then a Swiss chemist put a bell jar over a flower that was still growing and took nothing but the air inside it. The plant was unharmed. The analysis showed a scent profile that in many cases differed substantially from the extract everyone had been using for a century. Perfumery had been working from a translation without ever having seen the original.
A bell, a pump, and a tube of resin.
The apparatus is almost insultingly simple, which is usually the sign of a good idea. Enclose the subject in an inert glass or Teflon chamber. Draw the air out slowly with a pump, through a small cartridge packed with a polymer adsorbent, commonly Tenax or Porapak, which traps organic volatiles while letting air pass. Filter the incoming air so you are not collecting the neighbourhood. Run it for hours. Then take the cartridge to the laboratory, release what it caught by heat or by a tiny volume of solvent, and push it through a gas chromatograph coupled to a mass spectrometer.
What comes back is a list: every volatile compound the living plant was emitting, and in what proportion. From that list a perfumer can rebuild the smell on the bench, molecule by molecule.
The flower and the bottle are two different smells.
The general result, repeated across hundreds of species, is consistent in its direction even where the details vary. The living flower emits a profile weighted toward light, highly volatile molecules: green notes, small esters, terpenes, the fresh and slightly cool top that gives a garden at dusk its lift. Solvent extraction, which works on picked material over hours, preferentially recovers the heavier and less volatile end, concentrates it, and adds compounds formed during the process itself.
Neither is a lie. The absolute is a real material with real beauty, and in many cases it is richer and more usable than the living emission. But it is a portrait made after the sitter left the room.
Perfumery had been working from a translation for a century without ever having seen the original.
Some famous flowers have never existed as materials.
This is the part most people find genuinely startling. A number of the best-loved floral notes in perfumery have no natural extract at all, and never have. The flowers are what the trade calls mute: they yield nothing usable to distillation or solvent, or so little at such cost that no commercial material exists.
Lily of the valley
Muguet. Among the most reproduced notes in the entire industry, and there has never been a lily of the valley oil. Every muguet you have ever smelled, in every fine fragrance and every soap, is a construction.
Lilac, hyacinth, freesia
Spring in a bottle, and none of it extracted from the flower. Hyacinth absolute exists in tiny quantity and is nothing like the living plant. The others are built entirely from other materials.
Gardenia, peony, magnolia
Gardenia has no viable absolute at commercial scale. Peony has none. These are among the most requested notes in the market and they are, without exception, accords rather than ingredients.
Before headspace, these reconstructions were built from imagination and memory: a perfumer would smell the flower in a garden, go back to the bench, and try to approximate it from what was available. Some of the great muguet accords were made exactly that way and are triumphs of the ear rather than the instrument.
After headspace, the same job could begin from a list. You could take the bell to the plant, find out what it is actually emitting, discover the two molecules nobody had guessed at, and build from evidence. It did not remove the artistry. It removed the guessing about what you were aiming at.
A flower is not emitting the same thing at every hour.
Because headspace samples a living organism, it exposed something extraction could never show: emission is rhythmic. Plants release scent on a schedule tuned to the pollinators they are trying to reach, and both the quantity and the composition change through the day. Night-scented flowers ramp up after dark. Day-flowering species peak in warm morning sun. Some species change their blend as well as their volume, effectively saying different things at different hours.
This is why the jasmine harvest happens before dawn, a practice that predates any instrument by centuries. The pickers were right, and headspace explained why.
Roman Kaiser, and five hundred flowers that are disappearing.
The technique arrived in the late 1970s and through the 1980s, with work at several houses. At IFF a programme under Braja Mookherjee produced the trademarked Living Flower materials. At Givaudan, where he has worked since 1968, the Swiss chemist Roman Kaiser became the figure most associated with the method, and he took it somewhere nobody expected.
Kaiser used headspace as a form of botanical documentation. He carried the apparatus into rainforest canopies, up ropes into the crowns of trees, onto cliffs and into swamps, and analysed the scent of plants that were on their way out of existence. By the end of 2010 he had analysed the scent of some 520 endangered plant species, of which 267 appear in his book Scent of the Vanishing Flora.
Consider what that archive actually is. For a number of those species, a chromatogram in a Swiss laboratory is the only complete record that will exist of what they smelled like. The plant may go, and the description of its scent will remain, in a form precise enough to be rebuilt. It is a seed bank for smell.
We find that unbearably moving, and it is not a sentiment we can pretend is separate from the work of this house. A record of a scent made so that it can outlive the thing that made it is, exactly and literally, fragrance through memory.
You can put the bell over anything.
Once you have an instrument that samples the air around a subject, the subject does not have to be a plant, and this is where the technique becomes strange and a little unsettling.
Houses have taken headspace of a room, of the inside of a violin, of sea air, of the space above a glacier, of a market at a specific hour, of soil after rain. The technique is used to capture the smell of a place with the same precision it captures a blossom, and there is nothing in principle stopping you from taking the headspace of a person, a house, or a garment belonging to somebody who has died.
That last sentence is not hypothetical. The capability exists and is routine. Whether it should be used that way is a different question, and one that the industry has mostly not been asked in public.
Is a headspace reconstruction natural?
No. Not in any ordinary sense, and the marketing around this technology has been consistently slippery about it.
A headspace reconstruction contains, typically, no material from the plant whatsoever. It is a set of aroma chemicals, mostly synthetic, assembled to match an analysis. When a brand says it has captured the scent of a rare orchid, what has been captured is a data file. What is in the bottle was made in a factory. The claim is not false, but it invites a reading that is.
There is a real philosophical question underneath the marketing one, and we think it is genuinely open. A recording of a violin contains no wood, no gut and no rosin, and nobody says a recording is fraudulent. It is understood to be a different kind of object with its own honesty. A headspace reconstruction is, arguably, a recording in exactly that sense: a faithful transcription in a different medium.
The difference is that everyone knows what a recording is. Very few people know that the lily of the valley in their soap has never been near a flower, and the industry has had little incentive to explain it.
And there is a cost the technique cannot avoid. A reconstruction contains what the instrument could detect and identify. Anything present below the detection limit, or too unstable to survive the trap, or simply unidentified, is absent. Naturals are famously full of trace components in the parts-per-million that nobody has fully catalogued, and they matter. A headspace rebuild is frequently more accurate than an extract and slightly less alive, which is a sentence that could describe a great deal of modern technology.
For some of those plants, a chromatogram in a Swiss laboratory is the only record that will survive of what they smelled like.
The first technology for remembering a smell.
Writing preserved language. Photography preserved faces. Recording preserved music and voices. Smell, the sense most bound to memory, had no preservation technology at all: for the whole of human history, when a scent was gone it was gone, and the only copy was in somebody's head, degrading.
Headspace is the first real answer to that. It is not a perfect answer, for the reasons above. But it means that for the first time a smell can be written down accurately enough to be reproduced by somebody who never smelled it, in a hundred years, from the notation alone. That is the invention of the score, for the one sense that never had one.
A house named for the idea that fragrance carries memory should be clear about what that makes possible and what it does not. It does not let you keep the flower. It lets you keep the description, precisely enough that the flower can be spoken again. Which is, more or less, what memory does anyway.
The other two halves.
Nobody Can Predict a Smell
We can record a smell with an instrument. We still cannot look at a molecule and say what it smells like. Take the test, then meet the receptor code and the machine that predicts anyway.
The Vat and the Hillside
If a scent is information, it can be rebuilt without the plant. Engineered yeast now makes a patchouli molecule from sugar, and the rules let you call it natural.
Figures 2 and 3 are schematic. They illustrate directions of difference that are well documented across the literature rather than measurements of any particular sample.
- The technique, its 1980s development and the bell jar method: Headspace technology; Headspace technology in the perfume industry.
- Roman Kaiser, Givaudan, and the endangered species archive: Roman Kaiser; Scent of the Vanishing Flora, reviewed; Kaiser, Scent of the Vanishing Flora.
- Capturing a living flower without touching it: Headspace, capturing the scent of a living flower.
- Related reading in this journal: Jasmine, on the flower that cannot be distilled, and What a Formula Actually Looks Like, on what a reconstruction is written as.