The shape of a 2,211-material perfumery catalog

This is what one classified perfumery catalog contains, counted rather than described. Every figure in the tables below is computed from the material file PerfuMate ships, at the moment this page is built, so none of them was typed by anybody.

Read this as a claim about this catalog, not about perfumery. These 2,211 materials were chosen to be the ones a formulation tool needs to carry. Where a number below would be interesting as a general fact about perfumery, it is still only a fact about these rows. Nothing here is a survey.

What does a row actually carry?

A catalog is only as good as its thinnest column. These are counted at build time, so a field that stops being filled shows up here as a smaller number rather than as nothing.

Field	Materials	Share
A written odour description	2,211	100%
An olfactive family, not a filing bucket	2,154	97%
A sub-family	2,187	99%
A volatility band: top, heart or base	2,157	98%
An odour strength, 1 to 10	2,125	96%
A CAS number	1,642	74%
Supplier or trade synonyms	1,959	89%
An estimated constituent table	739	33%
Facet tags	2,209	99.9%

The CAS gap is worth stating plainly, and worth counting rather than characterising: 569 rows carry no CAS number, of which 35 are naturals and 534 are synthetics. The naturals are the expected part, because one molecule does not describe an essential oil. The synthetics are the larger group, and 418 of them carry a supplier or trade synonym in this catalog.

IFRA limits, GHS hazard data and the EU allergen list all join to a material by CAS, and a row's own CAS is not the only one it offers: 325 of those 569 rows carry an estimated constituent table, and all three checks join through the constituents as well. That leaves 244 rows, 11% of the catalog, with no CAS of their own and no breakdown to stand in for one. Those are the rows where an empty answer would look exactly like a clean one. The app's own under-assessed test is wider still at 307 rows, because a natural with a CAS but no breakdown hides what is inside it behind the oil's own number. PerfuMate marks all of them unknown rather than safe.

How much of this catalog is natural?

501 of 2,211 rows are naturals, 23%, and the split by family is nowhere near uniform.

Median strength is a filing, not a measurement. It is a perfumer's judgement of how much a material makes itself heard at the same dose, recorded on one 1 to 10 scale across all 2,211 rows. That makes the numbers comparable to each other and to nothing else. It appears in this table and the next one.

Family	Materials	Natural	Share natural	Median strength
Floral	451	85	19%	6
Woody	234	71	30%	6
Fruity	216	10	5%	6
Green	197	22	11%	7
Gourmand	179	32	18%	7
Herbal	146	68	47%	6
Citrus	144	56	39%	6
Spicy	102	56	55%	6
Resinous	81	41	51%	6
Musk	80	2	3%	5
Aldehydic	56	1	2%	7
Amber	53	5	9%	7
Earthy	42	16	38%	7
Animalic	41	6	15%	8
Sweet	31	2	6%	6
Aquatic	28	2	7%	7
Leather	21	1	5%	7
Smoky	16	6	38%	8
Ambergris	14	1	7%	7
Tropical	13	0	0%	6
Powdery	9	0	0%	6

Two more buckets exist and are left out of this table on purpose: 47 solvents and 10 rows filed only as other. 39 of them carry no olfactive family at all. The other 18 are filed in a bucket and still carry real families, and PerfuMate weights those normally: what its family weighting skips is the two bucket names, not the rows filed under them.

Do the top, heart and base bands mean anything measurable?

2,157 of the 2,211 materials are filed top, heart or base. Here is what those bands look like when you count them, and the answer to the heading is: not on loudness. The median odour strength is the same in all three.

Band	Materials	Share of catalog	Median strength	Share natural
Top	518	23%	6	28%
Heart	943	43%	6	20%
Base	696	31%	6	21%

54 rows sit outside the pyramid entirely. 47 of them are the solvents, every one: a diluent has no place in a scent pyramid, and giving it one would put weight where there is no smell. The other 7 are additives rather than odorants.

How loud is a typical material here?

This is the same filing as the two tables above, spread out: it is the number PerfuMate uses to work out impact rather than weight.

Strength	Materials
1	50
2	26
3	75
4	167
5	394
6	607
7	411
8	276
9	99
10	20
Not on file	86

119 materials sit at 9 or 10, 5% of the catalog. PerfuMate's AI does not dose the loudest bands like the rest: strengths 8, 9 and 10 each carry their own ceiling on how much of a concentrate they may be, which is a bracket per band rather than a figure per material. The 1,730 materials below strength 8 have none.

How many materials smell of more than one thing?

2,070 of 2,211, 94%, carry more than one olfactive family. That is the case against filing a material under a single heading, and it is why a scent profile here splits a material's character by dominance rather than assigning it whole to one family.

The most common sub-categories beneath those families. A material filed under the same sub-category through two different families is counted once here, so these are materials rather than mentions:

Sub-category	Materials
Leafy	389
Dry	384
Balsamic	280
Warm	265
Aromatic	243
Rose	230
Sugary	212
Green	168
Creamy	157
Herbal	151
Waxy	147
Honeyed	142
Jammy	136
Honey	134
Zesty	130

Where the data comes from, and what it is not

One perfumer, from supplier documentation, published literature and bench work, and it is the same file the app ships. Odour strength and the volatility band are judgements applied consistently across all 2,211 rows, so they are comparable to each other and to nobody else's scale. Constituent tables are estimates and say so. Where a material carries no CAS number, nothing was invented to fill the column.

Every material behind these numbers has its own page: start at the catalog, or a family such as floral or woody.