Data study · Zenyra
What perfumery stopped using
Four materials left the mainstream palette between the 1990s and today. They did not leave for the same reason, and the largest departure was not caused by regulation at all. Measured across 3,707 dated fragrances, of which 3,494 fall in the four decades shown.
1. Four materials left, for three different reasons
Perfumery’s palette is usually discussed as a matter of taste. The data says otherwise: the sharpest movements of the last thirty years are materials becoming unavailable, or uneconomic, rather than falling out of favour.
| note | 1990s | 2000s | 2010s | 2020s | change | cause |
|---|---|---|---|---|---|---|
| Oakmoss | 2.08% | 0.87% | 0.73% | 0.55% | −74% | Allergen restriction |
| Aldehydes | 0.49% | 0.41% | 0.23% | 0.16% | −67% | Fashion |
| Lavender | 2.41% | 0.87% | 1.24% | 1.25% | −48% | Fashion |
| Sandalwood | 4.48% | 2.85% | 2.86% | 2.69% | −40% | Supply collapse |
2. Oakmoss is the regulation story
Oakmoss falls 74% by share, the steepest movement in the dataset. Its restriction on allergen grounds is well documented, and the consequence is visible here: the base of the classic chypre stopped appearing in mainstream release. This is the case where cause and effect are both known and the number simply confirms them.
3. Sandalwood is not, and that is the more interesting finding
Sandalwood falls 40% by share, from 4.48% of all notes listed in the 1990s to 2.69% today. In absolute terms it was a far more common material than oakmoss to begin with, so more listings disappeared.
No allergen ruling explains it. Santalum album — Mysore sandalwood — was harvested faster than it could regenerate across the twentieth century, to the point of Indian state control of the trees and international protection of the species. The material did not become unsafe. It became scarce and expensive, and the industry moved to Australian sandalwood, synthetics, and reformulation.
The two curves also have different shapes, which is what makes them separable. Oakmoss collapses between the 1990s and 2000s and keeps falling as successive amendments tighten. Sandalwood drops once, between the 1990s and 2000s, then flattens at roughly 2.85% and stays there — the signature of a market finding a substitute and settling on it, not of a rule being progressively enforced.
4. The obvious objection: was it replaced rather than lost?
The first thing a perfumer will say is that sandalwood did not leave, it was substituted — Javanol, Ebanol, Sandalore and Polysantol do the work Mysore once did. That is true of the formulations. It is not visible in the data, and the reason is worth stating plainly.
Not one of those materials appears anywhere in the catalogue’s 5,308 entries. Nor does “santal”, “santalum”, or any other synonym — the catalogue says “sandalwood” or it says nothing. Brands do not disclose the molecule.
So this measures what was declared, not what was dosed, and the two possible readings pull in opposite directions. If houses reformulated to synthetics and kept writing “sandalwood” on the pyramid, the real decline in natural material is steeper than 40%. If they reformulated and stopped listing it at all, some of the fall is disclosure rather than composition. The honest position is that this page measures the second thing, and that the first is not recoverable from any note list.
What can be said either way: a material that appeared in 41.4% of 1990s releases now appears in 18.8% of them, and whatever replaced it is not being named.
5. Does excluding undated rows create the decline?
Roughly 30% of the catalogue has no verified release year and is excluded from the series, which is enough to matter if those rows were unusual. They are not, and the direction of the difference is the reassuring one.
| set | rows | containing sandalwood |
|---|---|---|
| Dated with notes | 3,717 | 784 · 21.1% |
| Undated, excluded | 1,589 | 256 · 16.1% |
| Whole catalogue | 5,306 | 1,040 · 19.6% |
The excluded rows contain less sandalwood than the included ones, and they concentrate in houses founded in the last fifteen years — Bond No. 9, Roja Parfums, Atelier Cologne, Memo Paris, BORNTOSTANDOUT. Dating them would add recent entries at close to today’s rate. It would not restore the 1990s baseline, and it would not reverse the finding.
6. The control group: what did not move
A decline is only meaningful against materials that held. These did.
| note | 1990s | 2000s | 2010s | 2020s | change |
|---|---|---|---|---|---|
| Musk | 4.54% | 5.19% | 5.24% | 4.78% | +5% |
| Rose | 5.74% | 4.66% | 5.28% | 5.36% | −7% |
| Jasmine | 3.77% | 2.75% | 3.29% | 3.19% | −15% |
Musk, rose and jasmine are as present now as they were thirty years ago. The palette was not thinned across the board — specific materials were removed while the base of mainstream perfumery stayed intact. Any account claiming the industry has been broadly stripped by regulation has to explain why the three commonest materials in perfumery did not move.
How these figures are measured
The obvious measure — what percentage of releases list a note — is unreliable across decades, because listing density is not constant. Entries from the 1990s carry 9.24 notes on average; entries from the 2010s carry 6.60. A material used exactly as much as before will appear to decline by roughly a quarter, purely because the denominator changed.
Every figure here is therefore a share of all notes listed in that decade, which is immune to that effect. A finding is reported only where the share moves by at least 35% relative. That threshold is a judgement rather than a statistical test, and is stated so it can be argued with.
This is not a hypothetical correction. Musk falls from 39.9% to 28.0% of releases and reads as a clear decline; by share it moves from 4.54% to 4.78%, meaning it did not decline at all. We published the uncorrected reading before running this check, and have since corrected it.
Limitations
Published in full, because a study you cannot check is not worth citing.
Note lists are marketing claims as much as formulation facts. They record what a brand chose to publish, not what a gas chromatograph would find. This matters most where a material has a marketing value out of proportion to its use.
31% of the catalogue has no verified release year and is excluded from every time series. The analysis rests on 3,707 dated fragrances with note data, and the 1990s column is the thinnest at 198 entries.
This is a curated catalogue, not a census. It over-represents fragrances people ask about and under-represents the very obscure, so it describes what was widely released and discussed rather than everything that existed.
No axis scores appear on this page. Warmth and sweetness are reliable, but projection, risk and texture are modelled for most rows and are not independent of each other, so they are kept out of a study that does not need them.
Figures are revised as the catalogue grows. Last recomputed 6 August 2026 against 3,707 dated fragrances. The catalogue is periodically extended, so a rerun months from now will return slightly different values; the direction and magnitude of every finding here has been stable across revisions. Cite the date alongside the figure if precision matters.
Cause is attributed, not measured. The dataset shows what disappeared and when. That oakmoss went for allergen reasons and sandalwood for supply reasons is drawn from the documented regulatory and trade record, not from the catalogue. The shapes of the two curves are consistent with those explanations, which is support rather than proof.
Use this
Published under a Creative Commons Attribution 4.0 licence. Reproduce any figure with credit and a link. The analysis is reproducible from the published catalogue via scripts/materials_study.py, and we will send the script to anyone who asks.
Companion study: what perfumery started using, and the broader Fragrance Taste Index.