Beginner perfumer · 99
What crosses over is edible: dual-use materials have a supplier median of 11, fragrance-only materials 4
· 12 min read
Last time I measured the two vocabularies in the uses field. This time I look at the materials themselves. Of the 7,807 records carrying the field, 34.5% have fragrance tags only, 15.2% flavour tags only, and 24.3% have both. The dual-use group is plainly better supplied — a supplier median of 11 against 4 for fragrance-only and 6 for flavour-only, and 11.2% of them have fifty or more suppliers against 2.6% and 1.6%. More interesting is which odour types cross over: caramellic 61.7%, waxy 56.9%, fatty 54.7%, green 53.1%, vanilla 51.2%, nutty 51.1% — all things you can eat. And at the bottom: amber at 11.6%, woody 22.5%, sulfurous 24.3%, floral 25.4%. Amber is not a food and not a plant. It is a concept perfumery invented.
About a 4 minute read.
Last time I measured the shape of the two vocabularies. This time I look at materials: who sits on both sides, and what that is worth.
Three piles first
Of the 7,807 records with potential_uses:
| Records | Share | |
|---|---|---|
| Fragrance (FR) tags only | 2,692 | 34.5% |
| Flavour (FL) tags only | 1,187 | 15.2% |
| Both | 1,900 | 24.3% |
| Tags unparseable | 2,028 | 26.0% |
Fragrance-only outnumbers flavour-only by more than two to one.
The dual-use group is plainly better supplied
| Median suppliers | Mean | Share with 50+ | |
|---|---|---|---|
| Both | 11 | 20.5 | 11.2% |
| Fragrance only | 4 | 9.2 | 2.6% |
| Flavour only | 6 | 9.9 | 1.6% |
A dual-use material is 4.3× more likely than a fragrance-only one, and 7× more likely than a flavour-only one, to have fifty or more suppliers.
Which is not hard to understand — something two industries buy has a different production volume and a different supply chain.
(Though mind the direction of causation: it may equally be that materials which were already commodity-scale get found by both sides. The data cannot separate those.)
And even within the dual group it is lopsided
Across those 1,900 records, the median flavour tag count is 2 and the median fragrance tag count is 10.
Even materials that work on both sides get five times more tags on the fragrance side. That matches the overall ratio from last time (FR occurrences are 6.5× FL).
Only 50 materials carry more than fifteen tags on both sides, and the top of that list clusters into two families:
| FL | FR | Suppliers | Material |
|---|---|---|---|
| 31 | 242 | 135 | Linalool |
| 31 | 237 | 15 | Coranol |
| 30 | 235 | 25 | l-linalool |
| 28 | 160 | 193 | Vanillin |
| 28 | 153 | 49 | Propenyl guaethol |
| 19 | 118 | 76 | Ethyl vanillin |
| 67 | 64 | 82 | Allyl hexanoate |
| 52 | 45 | 54 | Raspberry ketone methyl ether |
The linalool family and the vanillin family — which are precisely the best-supplied materials in the database. What works on both sides is what everybody already has.
(Note the last two: they are among the few where FL exceeds FR. Genuinely balanced dual-use materials are rare.)
Which odour types cross over
This is the most interesting table here. Dual-use rate by odour type (types with at least 40 records):
Highest:
| Dual rate | Odour type | Records |
|---|---|---|
| 61.7% | caramellic | 60 |
| 56.9% | waxy | 72 |
| 54.7% | fatty | 75 |
| 53.1% | green | 375 |
| 51.2% | vanilla | 43 |
| 51.1% | nutty | 92 |
| 49.5% | fruity | 521 |
Lowest:
| Dual rate | Odour type | Records |
|---|---|---|
| 11.6% | amber | 43 |
| 22.5% | woody | 293 |
| 24.3% | sulfurous | 152 |
| 25.4% | floral | 767 |
| 29.1% | balsamic | 203 |
Everything in the top group is something you can eat. Caramel, fat, nuts, vanilla, fruit — those words point at physical objects in a kitchen.
And amber sits at 11.6%.
Amber is not a food, not a plant, not even a raw material — it is a concept perfumery invented. (I have written about ambrein and amber naphthofuran; that accord is constructed jointly by a set of synthetics.) Nothing tastes of amber, so it does not cross.
Floral at 25.4% is worth a pause too — flowers are everywhere in perfume, and you rarely eat one.
And sulfurous at 24.3% is a different case: sulfur compounds are used heavily in flavour (I measured that 29 of the 0.01% evaluation tier are sulfurous), but most of them are flavour-only; they do not cross into perfume.
In one sentence
The direction of crossover is set by whether the thing has a physical referent.
Caramel, nuts and fat point at the same object in both worlds. Fougère, chypre and amber exist in only one of them.
In practice
- If you want cheap, easy-to-source materials, look at the dual group. A supplier median of 11 is nearly triple the fragrance-only figure.
- If you are building amber, woody or floral, do not expect cheap supply from the flavour side. All three sit below a 25% dual rate.
- The caramel, nut and fatty lines are worth approaching through flavour ingredients. Their dual rates exceed 50%, and flavour-side use-level specifications tend to be much clearer (the ppm ladders like buchu mercaptan's).
What this doesn't establish
- 26.0% of records had unparseable tags and were left out, so all three shares are underestimates.
- I cannot separate the direction of "dual means well supplied." Being dual may drive supply, or being widely supplied may attract tags from both sides.
- The dual rates by odour type use only records that have an
odor_type, and that field's coverage is not 100%. - "Amber is a concept perfumery invented" is my framing; the database does not classify it that way.
- Tag counts are not use volumes. 242 FR tags does not mean a material is actually used in 242 accords.
- I have smelled none of them.