Beginner perfumer · 103
Each order of magnitude of vapour pressure costs about a factor of three in substantivity — top, heart and base are computable from a spec sheet
· 12 min read
The last several articles have taken fields apart. This one does the opposite: it computes something usable out of vapour pressure. Taking the 1,288 records that have both a vapour pressure and a substantivity measured at 100%, the Spearman correlation is −0.647, and banding them is perfectly monotonic: below 0.0001 mmHg the median substantivity is 336 hours, 0.001–0.01 gives 142, 0.01–0.1 gives 48, 0.1–1 gives 12, and 1–10 gives 4. Roughly threefold per decade. Cut at 1 and 0.01 mmHg and the three bands line up with the traditional top, heart and base — ethyl acetate and cis-3-hexenol on top, geraniol and linalool and linalyl acetate in the heart, the lactones and methyl dihydrojasmonate and methyl cedryl ketone at the base. With one large caveat: 94.0% of vapour pressures are estimates.
About a 5 minute read.
The last several articles have been about taking fields apart. This one does the opposite: it computes something you can use from one field.
The question
"Is this a top note or a base note" is usually settled by experience or by somebody telling you. But evaporation rate is physics, and physics has numbers.
The database has a vapor_pressure field (9,020 records, 21.4%) and a substantivity field. If the first predicts the second, you can compute where a material sits from its spec sheet.
The correlation
I took records with both a vapour pressure and a substantivity measured at 100% — that restriction matters, because I have already been caught by the substantivity field's concentration marker, and a diluted figure cannot be compared to a neat one.
1,288 records qualify.
Spearman ρ = −0.647.
Negative, and not weakly so. Higher vapour pressure means shorter substantivity, exactly as physics predicts.
And banding it is perfectly monotonic
| Vapour pressure (mmHg @ 25 °C) | Records | Median substantivity |
|---|---|---|
| < 0.0001 | 49 | 336 hours |
| 0.0001 – 0.001 | 58 | 321 hours |
| 0.001 – 0.01 | 308 | 142 hours |
| 0.01 – 0.1 | 374 | 48 hours |
| 0.1 – 1 | 314 | 12 hours |
| 1 – 10 | 136 | 4 hours |
| > 10 | 49 | 4 hours |
From 336 hours down to 4, without once turning back.
The rough rule: each decade of vapour pressure is worth about a factor of three in substantivity.
(The top two and bottom two bands flatten — that is the substantivity field's ceiling and floor at work; I measured that 400 hours is a ceiling.)
Cutting it into three bands
| Band | Vapour pressure | Median substantivity | Representative materials |
|---|---|---|---|
| Top | > 1 mmHg | 4 hours | ethyl acetate, cis-3-hexenol, ethyl butyrate, benzaldehyde, ethyl 2-methyl butyrate |
| Heart | 0.01 – 1 | 24 hours | geraniol, citral, linalool, geranyl acetate, benzyl alcohol, citronellol, linalyl acetate |
| Base | < 0.01 | 212 hours | γ-undecalactone, γ-decalactone, δ-decalactone, methyl dihydrojasmonate, methyl cedryl ketone |
These three groups are the textbook top, heart and base. The heart list is especially clean — geraniol, linalool, citral, citronellol, the names any introductory book puts in the middle.
The odour types follow too: the top band is led by fruity (34), green (21) and ethereal (12); the base band by floral (79), fruity (26), woody (26) and balsamic (14).
A few checks
| Material | Vapour pressure | Substantivity |
|---|---|---|
| Ethanol | 44.6 mmHg | 1 hour |
| Linalool | 0.016 | 12 hours |
| Geraniol | 0.021 | 60 hours |
| Benzyl salicylate | 0.00017 | 384 hours |
| Iso E Super | 0.001 | 172 hours |
Note linalool against geraniol: near-identical vapour pressures (0.016 versus 0.021), and a fivefold difference in substantivity.
ρ = −0.647 is not 1.0, and that gap is what the remaining 0.353 looks like. Vapour pressure gives you a band, not an answer.
But there is a large caveat
8,479 of the 9,020 vapour pressures carry "(est)" — 94.0%.
That is the highest estimate share of any field I have measured. (For contrast, flash point is 56.6%.)
So the correlation above is really between an estimated vapour pressure and a measured substantivity. That it holds does not mean those estimates are accurate — it means the estimation model captured the same physical quantity.
Which arguably makes the result more useful: you do not need a measured vapour pressure to rank things. But do not quote it as a precise value.
In practice
- Meeting an unfamiliar material, look at the vapour pressure first. Above 1 is a top note, below 0.01 is a base note, in between is the heart.
- Do not rank within a band. Linalool and geraniol differ fivefold and vapour pressure cannot see it.
- Check first whether the substantivity figure was measured at 100%, or the two numbers cannot be put side by side.
- Only 21.4% of records have this field — where it is missing, boiling point is the next best proxy (same direction).
What this doesn't establish
- 94.0% of the vapour pressures are estimates, and I did not examine the estimation method. Everything here rests on that model.
- I used only the 1,288 records with substantivity at 100% — a subset of the 2,035 substantivity records, which are themselves 4.8% of the database. The sample is selected.
- ρ = −0.647 is a monotonic correlation, not a linear relationship. "Threefold per decade" is a rough description of the banded medians, not a fitted parameter.
- The top and bottom bands are flattened by the substantivity field's ceiling and floor, so the real spread at the extremes is wider than the table shows.
- The band thresholds (1 and 0.01 mmHg) are convenient cuts I read off the data, not any standard.
- I ran no blotter tests.