Notes from the Atelier: Four Formulas at Month Eighteen

This is a status report from the workbench, not a product announcement. At any given moment our perfumer in Grasse has several formulas in active revision, and four of them have been on the bench long enough that we want to write about them honestly. Two are close. One is experimental. One has resisted us for over a year.

None of them are on a launch calendar. We will say so when each is ready, and not before.

Why the work is slow

We do not ship a fragrance until we are right about it. That is the entire policy.

In commercial perfumery, a brief moves from conception to retail shelf in roughly nine to fifteen months. Many houses run faster than that. The cycle is shaped by sales calendars: a launch tied to a season, a campaign, a retail window. The formula has to be finished by a date that was set before anyone knew what the formula would be.

We work the other way around. The formula tells us when it is finished. If a composition is still mistranslated at month twelve, we keep working. If a base material we want is out of season, we wait for next year's harvest. If our perfumer says a particular accord is not yet right, we trust her. The bench has the casting vote.

Our perfumer works out of Grasse and has spent over a decade composing for niche houses. She is the person we send the brief to, the person who returns the trials, and the person who decides — with us — when a formula is honest. We do not name her publicly because that is her preference, and because the work matters more than the byline. What we will say is that her training is in the older French tradition: long maceration, high natural content, distrust of accords that arrive too easily.

The four scents below are at different stages. We are publishing this update because the people who follow this brand have been asking what is in development, and we would rather show the unfinished bench than invent a launch story. None of these are products yet. They are problems we are working on.

A small note on language. We sometimes get asked how many bottles each release will be. We make the kind of quantity a single workshop can produce by hand, with every bottle checked one by one, using materials whose supply is genuinely limited. The exact number is not the interesting part. The interesting part is that the work, by its structure, cannot scale.

Sake — month eighteen, still mistranslated

This is the formula that has resisted us the longest. The brief is to capture the smell of warm sake — the moment after pouring, when the cup is just at the lip and the steam carries the sweet, slightly lactic, faintly mineral note up to the nose. It is one of the most specific smells we know, and it is one of the hardest to render in alcohol.

The technical problem is straightforward, even if the solution is not. Sake's character on the nose comes from a shifting balance of esters, organic acids, and amino-derived volatiles produced by koji fermentation and yeast. Studies of sake aroma chemistry consistently identify isoamyl acetate (banana, fruit), ethyl caproate (apple, ginjo florals), diethyl succinate, and a handful of other esters as the dominant top notes. The lactic edge — the soft, almost yogurt-adjacent freshness — comes mostly from lactic acid, which makes up a significant share of sake's organic acid profile alongside succinic and malic.

The trouble is that almost none of those compounds survive isolation. Many of the key esters are highly volatile and below the threshold of perceptibility within minutes of pouring. The lactic and mineral character does not transfer cleanly into a perfume base because lactic acid is not, on its own, a fragrance material in the conventional sense. If you try to extract rice itself, the result is functionally silent on a blotter.

So we are reconstructing rather than extracting. The current direction uses a koji-fermented absolute we sourced through an artisan supplier in Japan, blended with a rice-husk tincture our perfumer macerated herself. We are testing a small dose of mate absolute in the warm phase — mate carries a bittersweet, slightly tobacco-adjacent character that, in trace quantities, reads as something close to the dry edge of warm rice. Tonka is in and out of the formula depending on the week; it adds the lactonic warmth we want, but in too high a dose it pulls the whole thing toward dessert, which is exactly where we are not going.

Where it stands: the warmth is right. The minerality is right. The transition between the two is wrong. There is a moment, around twenty minutes in, where the composition reads as idea of sake rather than sake. We are at month eighteen on this one, and we are still pulling at the edges.

Matcha — vegetal silence is harder than we thought

We assumed matcha would be the simplest of the four. We were wrong.

The misconception we keep running into — both in our own early trials and in the wider market — is that matcha smells like a tea-flavored drink. It does not. The smell of high-grade matcha is closer to wet stone, to the cool inside of a stone wall, to a field of grass that has been kept out of direct sun. There is sweetness, but it is the sweetness of stored chlorophyll and amino acids, not of fruit or sugar. Real matcha smells of restraint.

The chemistry confirms it. Matcha is shade-grown in the weeks before harvest, which suppresses photosynthesis and forces the plant to accumulate L-theanine and chlorophyll instead of converting them. L-theanine is the amino acid responsible for the deep umami quality on the tongue, and it travels into the aroma profile as a quiet, savory background. The volatile compound list is short and faint — terpenoids like linalool and geraniol show up in trace quantities, but matcha is, fundamentally, a low-volatility material. The leaves are speaking softly.

This is the texture we are trying to render, and it is the hardest texture in perfumery. Most synthetic green accords are built on materials that push the menthol, the cucumber, or the cut-grass facet. They produce something fresh, but they are loud where matcha is quiet. We are avoiding all of them.

The current direction in the heart is built on three materials. Oxalis contributes a sour-green facet that sits in the same family as wood sorrel — bright, slightly acidic, vegetal without being aggressive. Fig leaf, used at very low concentration, gives a milky-lactonic underlay; fig leaf has a coconut-adjacent sweetness from gamma-octalactone-like notes that, in small doses, reads as the soft creamy quality matcha gets in milk. (We are using a synthetic fig-leaf reconstruction here rather than the natural absolute, because the natural is phototoxic and IFRA-restricted in leave-on products.) Mate absolute sits heavier in the heart and gives the dry, slightly smoky tea-leaf body — mate is the closest natural we have found to a green-tea effect that is not synthetic.

Where it stands: this one is still translating. It feels like a careful description of matcha rather than matcha itself. Working.

Tattoo — the most experimental of the four

This started as a joke between us and our perfumer, and then we both stopped laughing.

The brief is the smell of fresh ink under skin — the iron in the pigment, the faint metallic warmth of healing, the salt of plasma at the edge of a fresh tattoo. It is the smell of being recently marked. It is intimate, and it is slightly unsettling, and that is exactly the point.

This is harder than it sounds because there is no tattoo accord in the perfumer's library. There is no reference material we can pull off a shelf. Every other formula in this article is, in some sense, a translation of an existing smell that already lives in nature or in a kitchen. This one we are building from scratch.

The current materials are doing four jobs. Cade oil, in rectified form, gives the tar-and-smoke quality that does the iron work. Cade is produced by destructive distillation of juniper wood and has an intense, smoky, leathery, phenolic character — it is the foundation of every classic leather accord in perfumery, and at low dose it reads as something between charred wood and warm metal. Dry tobacco absolute sits next to it for the cured-leather quality of skin that has been recently broken. Labdanum, in resinoid form, provides the sticky, animalic, slightly honeyed warmth that resin gives to amber and leather compositions; we are using the resinoid rather than the absolute here, specifically for its inkier, more leather-forward character. And a skin musk, sitting heavy in the base, crosses into the animalic range — the kind of musk that does not sit politely on the wrist.

The threshold question, which we are still testing, is the line between attraction and unease. A composition like this lives or dies on its dose. A drop too much of the cade and the whole thing reads as antiseptic. A drop too little of the musk and the metallic note sits on top, naked, with nothing to soften it. A drop too much of the labdanum and the experimental edge softens into a regular leather. We are walking a narrow line.

Where it stands: bench testing in low concentration only. This is the most experimental of the four, and the most likely to be a small, one-time release if it ever leaves the bench at all. We are not yet sure whether it should.

Sakura — closer than the rest, and still not the cliché

The first thing to say about this one is what it is not. The perfume of cherry blossoms in mass-market work is candy. It is sugar, vanilla, a fruity cherry note that has nothing to do with the actual flower, and a cosmetic-grade pink. We are not making that. We were never going to make that.

The actual smell of sakura is almost unfindable. Most species of ornamental cherry have a fragrance that sits at the edge of human perception — quiet, transparent, more sensed than smelled. The chemistry of cherry blossom that has been documented in perfumery research includes anisaldehyde (a powdery, slightly sweet floral), small amounts of coumarin (the hay-and-vanilla note that, in concentrated form, smells like cut grass on a warm day), and a thin green facet that prevents the whole thing from going syrupy.

What we are after, though, is not the moment of bloom. It is the moment after — when the petals have fallen, when the ground is faintly bitter with the green of bruised stems, when the sweetness has tipped just slightly toward decay. That is the smell most people associate with the season, even if they have never quite named it. It is a melancholy version of cherry, not a candy version.

The current formula leans on three things. Hawthorn carries the bittersweet floral fade — hawthorn has a strange dual character on the blotter, soft white-flower sweetness with an almost almond-anisic bitterness underneath, and it sits naturally in the territory of a fading petal. Pink pepper, derived from the berries of Schinus molle, gives the aerial spike at the top — pink pepper opens crisp, fruity, floral-spicy, and lifts the whole composition off the ground. And a single drop of nitrile musk in the base does the unsettling work; nitromusks have a sweet-powdery-animalic character that reads as warmth at a frequency the wearer notices but cannot quite place.

Where it stands: closer than the rest. This will be the first of the four to release, if it releases. The flower waits for no one — calendar pressure is real on a sakura formula, because the cultural reference itself is seasonal — but we will not push it out before it is right. If it is not ready for a given spring, it waits a year.

What happens next

We do not have launch dates for any of these. We are not going to manufacture one.

What we will do, when each formula is finished, is announce it first to the people on our list — the readers who chose to receive these letters in their inbox before anyone else hears anything from us. That is the small promise of being on the list. The four scents above will become products in the order they get there, on a timeline the bench dictates, not on a timeline we can show you in advance.

If you are reading this and want to know when any of these are ready, the way to be told first is the newsletter. Everything else is downstream of that.

— Sho