Perfumers are tired of being asked how long it lasts. The number exists — it just isn't measuring what the asker means.

A perfumer posts a question to a forum: what annoys you about this job? Fifty-seven replies arrive, and one complaint appears far more often than any other.

Being asked how long it lasts — before the person has smelled it.

Others put it differently: that anything under eight hours gets treated as cheap, that the conversation is only ever about projection and longevity and never about what the thing is doing. One reply blames a specific change: a large fragrance database moved from displaying a longevity time to rating it from excellent to poor, and a rating invites you to want more of it in a way a number does not. That excellent-to-poor line is drawn across a distribution with no natural cut points.

The irritation is fair. It is also worth noticing that the question has an answer, and that answering it properly turns out to explain the irritation better than dismissing it does.

What the data says, before any interpretation

This database records a longevity figure for 1,871 materials that have at least one supplier. Two things have to come out before the distribution means anything: 266 of those are censored rather than exact, recorded with a greater-than or less-than against thresholds from one hour to 168, and 246 were assessed below full strength or record no concentration at all. That leaves 1,372, and the distribution is not what the eight-hour argument assumes.

Longevity	Materials	Share
1 to under 4 hours	32	2.3%
4 to under 8	142	10.3%
8 to under 24	245	17.9%
24 to under 48	150	10.9%
48 hours and over	803	58.5%

The median is 76 hours. Eighty-seven per cent reach eight hours or more — the share strictly above eight is 83.5%.

If you had asked me to guess before I ran it, I would have said most materials cannot manage eight hours and the expectation is therefore unreasonable. That guess was wrong, and the correction is the interesting part.

So why does a finished perfume not last three days

Because those hours were not measured on anything resembling a wearer.

The source field does not record a bare number. It records, for example, 12 hour(s) at 100.00 % — the concentration is part of the measurement. Every row left in that table was assessed neat. Not diluted into a formula, not competing with thirty other materials for the same air, and not on a warm, moving, absorbent surface that is renewing its own oils all day.

A finished fragrance changes every one of those conditions at once:

• Concentration. A material present at 2% in a composition is not the material that was tested at 100%.
• Company. Longevity in a blend is a property of the blend. Materials suppress and extend each other, and the figure for one of them in isolation does not survive contact with the rest.
• Surface. Skin is warm, it perspires, it rubs against fabric, and it absorbs. None of those is true of the surface a laboratory figure is taken on.

Seventy-six hours describes a material neat, on a strip. Four hours describes a diluted perfume on a person’s skin. The two measure different subjects, so both can stand.

How much of each is actually documented

The dilution gap has no general multiplier. A material's share of a finished perfume is the concentrate strength times its proportion within the concentrate, and the second term spans orders of magnitude. RIFM's use-level surveys report 95th-percentile concentrations in fine fragrance per material: 0.0065% for linalool oxide, 0.68% for musk ketone, 3.0% for HHCB — the same percentile, hundreds of times apart. Concentrate strength does have a convention: IFRA gives EDT at 5–15% (typically around 10%) and EDP at 10–20% (typically around 15%), and states plainly that this is common industry practice rather than a classification standard.

Skin runs warmer than a strip, by a measurable amount. Typical forearm skin is around 31°C against 20–25°C indoors, a difference of 6 to 11 degrees. Vapour pressure rises with temperature per Clausius–Clapeyron, by an amount that depends on the molecule: limonene gains about 41% from 20°C to 25°C, linalool oxide about 69%. That gives the direction and not a conversion to hours, because adsorption, penetration, humidity, airflow and the base all act at the same time.

Skin absorbs, and it has been quantified. In non-occluded in-vitro human skin, linalool absorption over 24 hours ran about 1.84–4.08% of the applied dose, rising to 5.9–14.7% under occlusion. Skin also metabolises some materials — benzyl acetate is hydrolysed by esterases. Absorption does not make everything vanish sooner: 2025 work found skin retains some molecules while others release more from skin than from glass.

A correction about the 76 hours itself

We wrote this treating it as a measurement of whether the molecule is still present. That is wrong. TGSC lists values like "172 hour(s) at 100.00 %" under Organoleptic Properties / Substantivity, sourced to Luebke 1989. It is a sensory detection time, not GC, mass spectrometry or residual mass.

So both numbers measure the same kind of thing — whether it can be smelled — under different conditions: a pure material on a strip against a database's sensory endpoint, versus a finished product on skin at a real dose with the endpoint decided by the wearer, usually without a stated criterion.

TGSC's public pages also do not state the strip material, applied mass, number of assessors, sniffing interval, detection criterion or uncertainty.

And the field's resolution is one hour

Perfumers discussing longevity usually start at half-hour intervals, because the early part of the evaluation is where the top notes move. This field cannot record that.

Across the 2,027 rows carrying an hour figure:

Below one hour	0
Carrying any decimal	0
Sitting on 1, 2, 4, 8, 12, 24, 48, 72, 100, 168, 200 or 400	1,027 (51%)
Eight hours or less	408
Two hours or less	88

Half the dataset sits on twelve round values, and the smallest step is an hour. So across the range people actually argue about — whether the top is still there — the field offers four positions, 1 through 4, and two of those are among its favourite landing places.

That is the same point as the missing measurement conditions, seen from the other side: one is about what was not written down, the other about what the format cannot hold. The 76-hour median should not be read as a precise standard value; it is the median of a consistently sourced set whose conditions are undocumented.

Olfactory adaptation does not carry the explanation

There is a tidy account available: you stop smelling your own perfume, so "five hours" measures your nose rather than the perfume. Checking it, the account cannot bear that much weight.

Short-term studies do show rapid decline — olfactory event-related potential amplitude falls about 48% under repeated stimulation with a time constant of roughly 4–10 seconds, and in ten-minute continuous exposure perception peaks near 10 seconds and reaches its minimum around 120 seconds. A four-hour room study found the low-concentration group close to the odourless condition after about two hours, and two weeks of domestic exposure at six hours a day produces durable, odour-specific threshold elevation.

None of those is the same situation. No study we found measures the wearer ceasing to perceive at hour five while instruments or third parties continue to detect, and skin concentration keeps falling, airflow is intermittent and the accord's composition changes throughout, so constant-stimulus results do not transfer. We also could not find peer-reviewed work comparing wearer and third-party judgements of the same endpoint.

Which makes the question answerable, rather than ignorant

The person asking how long it lasts is asking about their own experience — how long they will smell of this. That is a legitimate thing to want to know, and it is not what a materials database is reporting.

One question back settles it: on what? A strip, your skin, this concentration, this climate — four conditions give four numbers, and the one they meant is usually the skin one.

That reframing also disposes of the eight-hour standard. Eight hours on skin is a demanding target that says something about the formula. Eight hours neat is unremarkable — 87% of the raw materials left in that table clear it without trying, including many that vanish from a wearer within the afternoon.

We publish this number too, so this is our problem as well

Every material page on this site prints a longevity figure. It comes from the same field, with the same measurement conditions, and if we display it without saying what it refers to, we are producing exactly the confusion the perfumers in that thread are tired of.

That is the reason for this article. A number without its conditions is not information; it is a suggestion. The database is more useful when it says what it measured than when it produces a figure that looks like an answer to a question nobody asked it.

What this does not establish

• The figures are as recorded by the source, not measured by us. They carry the usual variation of any panel assessment. The censored entries excluded above are 14% of the raw set, and dropping them removes information rather than noise: a material recorded as > 168 hour(s) is genuinely long-lasting and is now absent from the count. Opening those censored strings is worse still: the 400 hours quoted for Ambroxan is a floor measured at one tenth strength in dipropylene glycol, and the numeric column discards all three facts.
• Nothing here predicts skin longevity from the material figure. The relationship exists but is loose; what a boiling point can and cannot substitute for works through a related case where a ranking survives and a prediction does not.
• The complaint is one thread on one forum. It is unusually consistent for a thread of that size, but it is not a survey.