Starting out · 5
The database has 42,225 entries. Your actual palette is about three thousand.
· 11 min read
Browsing a materials database feels like unlimited choice. It is not. Of 42,225 entries here, 6,913 have a described smell, 4,462 of those have three or more suppliers, and 134 odour families collapse into ten that cover two thirds of everything. What that means for someone deciding what to buy.
Open a raw materials database and the first feeling is that there is too much. Forty-two thousand entries, most of them names you have never seen. It reads as unlimited choice, and the natural response is paralysis.
The choice is not unlimited, and the number that matters is much smaller. Here is the whole corpus narrowed step by step, each step for a reason.
Step one: does anyone say what it smells like?
42,225 entries → 6,913 with a described smell.
Only 16% of the database has an odour description at all. The rest are there for other reasons — flavour materials, solvents, intermediates, things listed for regulatory identification. The datasheet article makes the same point from a different direction: not every row in a fragrance database is there to be smelled.
So five sixths of that intimidating number was never a candidate.
Step two: can you buy it?
6,913 described → 5,824 with at least one supplier.
Across the whole database, 49.1% of entries have no supplier listed at all. You can look them up, read their properties, and not obtain them. Among the ones with a described smell the picture is much better — 84% have at least one supplier — which makes sense, because a material nobody sells is a material nobody bothered to describe.
Tighten it to materials with a real chance of being in stock somewhere:
| Filter | Materials |
|---|---|
| Described smell | 6,913 |
| …and ≥1 supplier | 5,824 |
| …and ≥3 suppliers | 4,462 |
| …and ≥5 suppliers | 3,372 |
| …and ≥10 suppliers | 2,073 |
An important caveat about that column. These are suppliers listed against a material in the source data, which mostly means industrial suppliers selling in industrial quantities. A material with twenty suppliers may still be impossible to buy in 10 g as a hobbyist. Read the count as "is this a real article of commerce or a curiosity", not as "can I get this next week". For what you can actually order in small amounts, the constraint is the handful of retailers who repack — a much shorter list than any of these numbers.
Step three: the families collapse
Among the 4,462 materials that are both described and reasonably supplied, the source assigns 134 different odour families. That still sounds like a lot.
It is not, because they are wildly uneven:
| Family | Share |
|---|---|
| floral | 14.4% |
| fruity | 11.6% |
| herbal | 7.9% |
| green | 7.7% |
| citrus | 6.1% |
| spicy | 6.0% |
| woody | 5.5% |
| balsamic | 3.8% |
| sulfurous | 3.5% |
| nutty | 1.9% |
The top ten families cover 68.3% of the buyable, described palette. The remaining 124 families share the other third between them.
What this actually changes
The constraint on a beginner is not imagination, it is availability — and availability is concentrated. You are not choosing from forty-two thousand things. You are choosing from a few thousand, most of which sit in ten buckets.
That is good news in a specific way. It means:
- A palette of ten well-chosen materials is not a token gesture. Against a real working set of a few thousand, ten covering several of the dominant families is a genuine spread. This is the arithmetic behind buying ten materials rather than a thirty-vial kit.
- The gaps in your palette are predictable. If you have nothing herbal and nothing green, you are missing about 15% of the accessible palette, and no amount of buying more florals fixes it.
- Rarity is usually a supply fact, not a quality fact. A material with one supplier is not more special than one with thirty; it is harder to get, which is a different property and often a worse one when you are learning and want to repeat an experiment.
And what the fragrance wheel does not tell you
The familiar fragrance wheel arranges finished perfumes, so it cannot show you this. It has no way to say that half the entries in a materials database cannot be bought, or that the material families are this lopsided.
Those two facts are the shape of the palette you will actually work with. The wheel describes the destination; this describes the shop.
How to filter, practically
When you are looking at any materials database, apply the three steps in order before you get attached to anything:
- Is there an odour description? No description means nobody in the trade treats it as a fragrance material.
- How many suppliers? Below three, treat it as an interesting entry rather than a purchase candidate, and check a repacker before planning around it.
- Which family, and do you already have that family covered? Adding a fourth citrus is a smaller gain than adding a first herbal.
Then, before you order, read the datasheet properly — the flash point in particular, since it decides whether the thing can be shipped to you at all.
And if you want to see what a worked-through palette looks like rather than a filtering rule, the starter palette is mapped out material by material — 46 of them, each with a one-line role.