Top, heart and base are not three stages — evaporation is one continuous curve

Top, heart and base are the first three words a beginner learns. Useful words; every teaching text leans on them. The bottle, though, contains no such stages.

What happens in the bottle is this: every molecule starts evaporating in the same second. They just do it at different speeds. "Top note" is not a stage; it is the period during which the fast molecules are still in the air.

That sounds like a quibble until you lay a formula out by the numbers.

How the research measures it

In 1995 Vuilleumier, Flament and Sauvegrain published a dynamic headspace method: a collection device on the inner forearm draws diffusing organic vapours from the skin onto a Tenax adsorbent at a controlled air flow rate, and the concentration of each component in the gas phase is measured against time.

They tested it with a model composition of eleven synthetic odorants in an alcoholic matrix, and what came out was one decay curve per component. Some fall below detection within minutes. Others are still there hours later. The speeds spread out continuously, with nothing in the data that looks like three stages.

So "top, heart, base" are lines we draw across those curves afterwards. Where the lines go is a convention, and textbooks disagree on where the boundaries sit.

And the curve differs between people

In 2025 Hadjiefstathiou and colleagues did something harder: an in vivo study across several volunteers, using GC-FID to semi-quantify the evaporation of eight fragrance molecules (myrcene, limonene, ethyl heptanoate, ethyl octanoate, citronellol, hexyl cinnamaldehyde, ethyl decanoate, ethyl butyrate) from the forearm, with glass surfaces as controls. They measured each volunteer's skin parameters as well.

The conclusion: skin properties do influence how fast fragrance molecules evaporate, and the variation between individuals is clear. Statistical analysis indicated that the physicochemical properties of the molecules and the skin type together account for the differences.

Every perfumer knows the same bottle smells different on different people. Few have ever had numbers for it. Now there are some: the difference is real and measurable, skin itself is part of the cause, and body odour alone will not explain it.

Substantivity in our data

Of the materials we have compiled, 1,871 carry both a substantivity figure and supply information as of August 2026 — the count moves as the source is re-read, and it was 1,838 when this article was written. Before you use the table, know where these numbers come from:

These are residence times on a smelling strip, not on skin, and the source rounds them coarsely — 43.5% of values sit on round numbers like 4, 8, 12, 24, 48, 168 and 400 hours, with "400 hours" alone accounting for 306 entries.

Treat them as an ordering scale, not as absolute values. Putting a boiling point beside them shows what that limitation costs: the ranking holds across every band, and the spread inside one band is wider than the gap between neighbours.

With that caveat, the median substantivity by odour family (families with 40 or more materials):

Odour family	Median substantivity	n
Minty	16 hours	41
Fruity	18	184
Green	21	187
Citrus	28	118
Herbal	36	133
Floral	88	344
Spicy	144	87
Woody	284	131
Balsamic	344	99

A 21-fold spread from minty to balsamic, and it is continuous. Nowhere in that table is a natural break where you could point and say "above this line is base".

This explains the advice in the last piece

The opening move from the previous piece was: one floral in the middle, one lighter, one heavier.

Floral's median is 88 hours, between citrus (28) and woody (284). Citrus is fading while the floral is still there; the floral is fading while the wood has already taken over. The floral is the joint in odour terms and the joint in time as well.

Part of the reason "floral × citrus" and "woody × floral" come up as such frequent pairings is that they sit next to each other on the evaporation curve.

Building a curve that does not collapse

In practice the job is one line, kept covered end to end.

1. Lay your materials out by substantivity and look for the empty stretch. My own first order was the classic mistake: five citruses and one cedarwood, because citrus was cheap and smelled good straight away. The blend "disappeared" twenty minutes after spraying, and two hours later my wrist smelled of wood and nothing else. Only when I lined the numbers up did I see I owned nothing between 12 and 48 hours. Most beginners' gap sits right there.

2. Keep neighbouring materials roughly 2–3× apart in substantivity. A 10× jump is a hole. Citrus at 28 hours straight into wood at 284 is exactly that jump, and something floral or herbal has to fill it.

3. Add the slow material early. The classic way to ruin a blend is to perfect the top first, drop the base in last, and watch the whole thing change. Base molecules evaporate from the first second; they alter every phase you smell, including the opening you thought was finished.

4. On the strip, write times: fresh, 10 minutes, 1 hour, next day. "Nice" or "not nice" tells you little a week later. You are listening for the moment the strip goes empty, because that hole is where the next material goes.

One idea to drop

"The longer the base lasts, the better."

Many of the materials above 240 hours are musks and balsams. They last; holding the middle is a separate job. A fragrance made of citrus and musk has a hollow middle, and the middle is exactly when other people smell you.

Substantivity is one axis to sort along. Treat it as a score and the formula drifts toward the heavy end.

References

C. Vuilleumier, I. Flament & P. Sauvegrain, Headspace analysis study of evaporation rate of perfume ingredients applied onto skin, International Journal of Cosmetic Science, 17, 61–76 (1995). doi:10.1111/j.1467-2494.1995.tb00110.x

E. Hadjiefstathiou, G. Savary, C. Malhiac, D. Terescenco & C. Picard, Exploring the impact of fragrance molecular and skin properties on the evaporation profile of fragrances, International Journal of Cosmetic Science, 47(6), 981–995 (2025). PMID 40524649

---

Next up is IFRA: what the limits actually restrict, why they so often change the structure of a formula, and when a beginner genuinely needs to care.