Detection Threshold Calculator
Combine a panel’s individual threshold estimates into a single group threshold as the geometric mean, the way EN 13725 specifies.
Group threshold
EN 13725 combines panel results as the geometric mean of the individual threshold estimates — the antilog of the mean of the log₁₀ values — because olfactory thresholds are log-normally distributed. A plain arithmetic average would overweight the least-sensitive panellists.
Why the maths is logarithmic
People differ enormously in how faint an odour they can detect, and that spread is multiplicative, not additive — one assessor might need ten times more odorant than another. When data behave that way, the geometric mean is the honest measure of the middle, and it is what olfactometry standards call for.
The calculator averages in log space and reports both the group threshold and the mean of the logs, so you can see the working. It is the same reduction step a laboratory applies after a session — but a real result also depends on panel selection, screening, and instrument calibration that no formula can supply.
Frequently Asked Questions
What is an individual threshold estimate?
It is the dilution factor at which one panel member's response flips from 'no odour' to 'odour' — the point where that person just detects the sample. Each assessor produces their own estimate on each presentation.
Why use the geometric mean instead of a plain average?
Olfactory sensitivity is log-normally distributed across people, so threshold data span a wide multiplicative range. The geometric mean — the antilog of the mean of the log10 values — is the statistically appropriate centre; a plain arithmetic mean would be dragged up by the least-sensitive assessors.
How is the geometric mean computed here?
The tool takes log10 of each positive estimate, averages those logs, then raises 10 to that average. For estimates of 10 and 1000 the logs are 1 and 3, the mean is 2, and 10^2 = 100 — the group threshold.
How many panellists do I need?
EN 13725 sets minimum panel sizes and screens members against a reference gas (n-butanol) so the panel's sensitivity is known. This tool will average whatever positive estimates you enter, but a defensible threshold requires a properly selected and screened panel.
Can I mix retained and rejected data?
No — enter only the valid individual threshold estimates you intend to keep. Quality screening (removing physiological outliers, checking repeatability) happens before this averaging step, following the standard's rules.
Screening estimates only — not a substitute for accredited EN 13725 olfactometry, dispersion modelling, or regulatory advice.