How we source

Testing and measurement

We do not publish wear tests. We may use a bottle to check a detail, but the impressions on this site are not presented as controlled test results. If we publish a test in future, the page will describe the method and results.

Descriptions of longevity and projection are impressions, not measurements. For entries based on reader reports, we describe the reported range rather than imply a precise duration or distance.

Where the facts come from

We cite published sources for the notes, the perfumer, the launch year and other product details, using the house's own pages where available. Formulation records may also cite regulators or community reports; the page identifies the source and the kind of evidence it provides. Sources are listed at the foot of each fragrance page.

A published note list describes the scent as the house presents it. It is not an ingredient list or a formula. When we cannot find a published note list, we say so and leave it out.

We do not use other fragrance databases as the source for a house's note list.

Where the impressions come from

An impression describes how a fragrance seems to wear. It is labelled separately from the cited product details. Our threshold for a corroborated summary is:

• At least ten independent reports from at least two separate venues.
• Reports are counted individually. Replies that only repeat another person's account do not count as independent reports.
• Findings are described as reports, not measurements: for example, "Buyers widely report it wears close to the skin."
• Summaries use our own words and do not identify private individuals.

Where this site stands today. Of the 96 impressions currently published, 86 clear that floor. 6 are drawn from counted reports that fall short of it, and each of them says so on its own page, with the report and venue counts. 4 older impressions are not backed by counted reports. They are being replaced.

An impression based on fewer reports or fewer venues is labelled as falling below the threshold. Its page shows the report and venue counts. Older impressions without counted evidence are labelled separately, as are entries written by a named editor.

When reports disagree, we describe the disagreement. A "contested" label means the reports differ; it does not establish that a formulation changed.

The current bottle

Fragrance formulas can change. Each fragrance page includes a formulation status. "No published reformulation record" means we have not found evidence that establishes a change; it does not prove the formula has stayed the same.

We distinguish a change documented by a house or regulator from one reported by buyers. The wording and citations on the page explain which evidence supports the status.

When we get it wrong

Report an error to [email protected]. We note corrections on the affected page and cite the supporting source. The editors reserve the right to correct a factual error brought to their attention; any such correction must be attributable to a reliable source.