How a fragrance is built: accords, structure and why blends smell muddy

An accord is a small group of materials — often three or four — that together smell like a single thing which none of them smells like alone. It is the perfumer’s equivalent of a chord, and the analogy is close enough to be useful: the point is the combination, and the individual notes stop being separately audible.

This matters because it is how composition actually proceeds. A perfumer does not add thirty materials to a beaker; they build two or three accords, each of which is a coherent smell, and then compose those. It is the difference between writing with words and writing with letters.

The practical consequence for a beginner is that the target is not a long ingredient list. A blend of five materials you understand will nearly always be better than one of twenty you do not, and it has the enormous advantage of being diagnosable — when it is wrong, you can tell which one is doing it.

Fragrance is conventionally described as top, heart and base notes, and the model is a rough description of volatility rather than a recipe. Volatile materials leave first, so what you smell in the opening minutes is not what you smell an hour later; heavier materials remain and dominate the long dry-down. Citrus and light aromatics are characteristically top; florals and spices are characteristically heart; woods, resins and musks are characteristically base.

What the pyramid does not mean is that a fragrance is three separate layers stacked up. The materials are all present together from the first moment and they evaporate at overlapping rates, so the experience is a continuous transition rather than three acts. A composition planned as three unrelated tiers usually smells like exactly that — a pleasant opening followed by something unrelated.

The structural skill is making the transition coherent: choosing materials whose character carries through as the volatile ones leave, so that a fragrance thirty minutes in is recognisably related to the one you first smelled. Materials that bridge — present enough to be noticed early and persistent enough to survive — are what do that work, and finding them in your own palette is a large part of learning the craft.

1. 1

   Step 1: Learn materials one at a time, on blotters, over months

   Smell a single material on a blotter, repeatedly, over hours — because it changes, and how it changes is most of what you need to know about it. Write down what it reminds you of in your own words rather than the supplier’s. This stage cannot be shortened and it is the whole foundation.

   You will know it worked when: You can name a material blind from your own palette.

2. 2

   Step 2: Build two-material pairs before anything larger

   Take two materials you know and smell them together at different ratios. This is where you learn that combination is not addition — some pairs produce a third smell and some cancel each other out.

   You will know it worked when: You have found at least one pair that surprised you.

3. 3

   Step 3: Build one accord and stop there

   Three or four materials aiming at a single coherent smell. Do not add a fifth because the fourth did not work; change the fourth. An accord that works is a building block you will reuse for years.

   You will know it worked when: A blend that smells like one thing rather than several.

4. 4

   Step 4: Change one variable at a time and write everything down

   The discipline that separates the craft from a series of accidents. Every trial gets a number, a full formula by weight, and a date. Changing two things at once means learning nothing from either.

   You will know it worked when: You can reconstruct any trial exactly from your notes.

5. 5

   Step 5: Let trials rest, and smell them again

   Blends change over days and weeks as materials interact and settle, and a judgement made in the first hour is frequently wrong. Label, wait, and revisit — the most common regret in the craft is discarding something that needed a fortnight.

   You will know it worked when: You have judged the same trial twice, days apart.