All posts

Reading a PFAS result: PFAS-4, sum of 20, detection limit, ng/L and µg/L

Published 13 September 2026 · pfasnearme

A PFAS result is a number, a unit, a parameter code, a date, a sampling point and often a "less than" sign. Each of those can change the meaning of the whole. This guide explains what every field on a commune page means, and why we display "no PFAS detected, under 4 ng/L" where others would display "compliant".

Where the number comes from

Everything you read on a commune page comes from the sanitary control of the regional health agencies, the ARS. An accredited laboratory samples water at a tap on the network, analyses it and publishes the result in the national database, which the Hub'eau API makes available to everyone [1]. We measure nothing ourselves. We read, convert, compare, and update weekly.

Since 1 January 2026, testing for 20 PFAS is mandatory in that control, for every commune [5]. Before that it depended on the ARS and the local context. That is why some communes have three years of history and others a single sample.

The unit, first

Laboratories publish in micrograms per litre, µg/L. We display nanograms per litre, ng/L. A microgram is a thousand nanograms. The European limit of 0.1 µg/L becomes 100 ng/L. The Swedish limit of 0.004 µg/L becomes 4 ng/L. Our benchmark of 0.0005 µg/L becomes 0.5 ng/L.

We make that conversion because decimals deceive. "0.012 µg/L" looks like nothing. "12 ng/L" is three times the Swedish limit. Same number, different reading.

The two sums that matter

The sum of 20 PFAS carries Sandre code 8847 [2]. It is the regulatory parameter, the one with the 100 ng/L limit. The 20 substances are perfluorinated acids of 4 to 13 carbons, ten carboxylic and ten sulfonic, listed in the December 2025 decree [5].

The sum of 4 PFAS, PFOA, PFOS, PFNA and PFHxS, carries code 9268 [3]. These are the four EFSA assessed together and for which it set its tolerable weekly intake [11]. This is the sum we compare with our 0.5 ng/L benchmark, and the HCSP compares with 20 ng/L [7]. The parameter has only existed in the national reference list since 25 April 2025 [3]. For earlier samples we add the four substances ourselves when they are published separately.

The trap is that the sum of 20 can be high with a low sum of 4, and the reverse. Short-chain PFAS, PFBA or PFHxA, inflate the sum of 20 without weighing in the four. So a commune can be "compliant" at 100 ng/L and far above our benchmark, or the other way round. Look at both. We explain why countries do not pick the same sum in the comparison of limits.

The "less than" sign, and why it changes everything

When you see "<0,004", the laboratory is saying it could not quantify the substance and that its limit of quantification was 0.004 µg/L, which is 4 ng/L. That is not zero. The true value is between 0 and 4 ng/L. Nobody knows where.

The European directive only requires the limit of quantification to be at most 30% of the parametric value [4]. For the sum of 20 at 100 ng/L, a laboratory is compliant with a limit at 30 ng/L. In practice French laboratories do much better, typically 1 to 5 ng/L per substance. Anses, in its national campaign, quantified PFOA from 2 ng/L [6]. But "much better" is still above 0.5 ng/L.

Laboratories, in France as in Germany, do not count values below the limit in sums [8]. Water with four substances all at "<4 ng/L" has a published sum of 4 equal to zero. Its true sum could be up to 16 ng/L. That is why our pages show "no PFAS detected" with the bound "< 16 ng/L", not zero. When every component is a non-detect and the limit is above our benchmark, we refuse to say "below 0.5 ng/L". We say what the lab looked for, what it did not find and what it could not have seen. Others would display "compliant". Both are true. The second is less useful.

Conversely, a sum built from non-detects counted as zero is a floor. It can prove an exceedance, never a pass. That is why "at least" appears on some averages.

The date and the sampling point

A commune can be supplied by several networks, called distribution units, with water of different origins. Lyon has several. When that is the case we name the network and build the history on the most frequently sampled one. If your address is on another network, your water may differ. Your mairie knows which one supplies you.

The date matters as much as the number. A 2023 result in a commune where a well has since been disconnected says nothing about today's water. We show the date of the latest sample and the date of our latest update separately, because they are two different things.

One sample or a trend

PFAS vary from one sample to the next. The laboratory changes, the season changes, the blend of wells changes. The HCSP asks to confirm an exceedance on the median of at least four months of monitoring over two seasons, and to apply a 30% expanded uncertainty to every measurement [7]. In the US, compliance is judged on a running annual average, never on one result [9].

We do the same. The history section of every commune shows all samples since 2023, and the summary gives the mean of the last twelve months. We only write "rising" or "falling" with at least three quantified values and a gap of at least 25% and 1 ng/L between start and end. One number going up is not a trend. And if the latest sample is a non-detect, we never say "rising", whatever the earlier ones show.

Why an average and not a peak

Because that is how the body counts. PFOA takes 2.7 years to halve in blood, PFOS 3.4 years, PFHxS 5.3 years [10]. What you drink is smoothed over years. A peak at 8 ng/L one month then 1 ng/L the rest of the year weighs less than water steady at 3. EFSA's weekly intake is a long-term budget, and that is the budget we compare against [11].

The thirty-second read

• Read the sum of 4 PFAS in ng/L. That is the health number.
• Check for a "less than" sign. If it is everywhere, the result is a floor, not a measurement.
• Check the date and the network. An old number or another network does not describe your tap.
• Open the history. Three values are worth more than one.
• Then compare with 0.5, 20 and 100 ng/L, and read what to do if you are above.

Frequently asked questions

What does "<0,004" mean on a PFAS result?

That the laboratory could not quantify the substance, and that its limit of quantification was 0.004 µg/L, which is 4 ng/L. It is not zero. The true value is somewhere between 0 and 4 ng/L. We display it as "not detected" and count it as zero in sums, which makes those sums a floor.

What is the difference between ng/L and µg/L?

A microgram is a thousand nanograms. The European limit of 0.1 µg/L is 100 ng/L. Our 0.5 ng/L benchmark is 0.0005 µg/L. Laboratories and Hub'eau publish in µg/L. We convert everything to ng/L because strict limits read better that way.

Why is the sum of 4 PFAS missing for some communes?

The "sum of 4 PFAS" parameter has only existed in the Sandre reference list since April 2025. Before that, laboratories published the 20 substances separately and we add the four ourselves. When they only published the sum of 20, we cannot isolate the four and we say so.

Why does my commune show "no PFAS detected" with "< 4 ng/L" rather than zero?

Because every component of the sum is below the lab's detection limit, and that limit is higher than our benchmark. Water at "<1 ng/L" for each of the four substances could be at 0 or at 4 ng/L for their sum. So we show the upper bound, "< 4 ng/L", and say how many times our 0.5 ng/L benchmark the lab could have missed. We refuse to declare it below 0.5 ng/L without proof.

How do you compute the trend?

We take the mean of the last twelve months on the commune's main network, and we only say rising or falling with at least three quantified values and a change of at least 25% and 1 ng/L. If the latest sample is a non-detect, we never say "rising".

Sources

1. Hub'eau: API Qualité de l'eau potable (the public dataset behind every commune page, from the ARS sanitary control)
2. Sandre: paramètre 8847, somme des 20 PFAS
3. Sandre: paramètre 9268, somme de 4 PFAS (PFOS, PFOA, PFNA, PFHxS), created 25 April 2025
4. Directive (EU) 2020/2184, Annex III Part B: laboratories must reach a limit of quantification of 30% or less of the parametric value
5. Décret n° 2025-1287 du 22 décembre 2025: the 20 PFAS in the French sanitary control, plus TFA and 6:2 FTSA from 2027
6. Anses (2025): Campagne nationale de mesure de l'occurrence de composés émergents dans les EDCH, PFAS et US-PFAS (per-compound limits of quantification, table 2)
7. HCSP (9 July 2024): avis PFAS dans les EDCH (20 ng/L for the sum of four, 30% expanded uncertainty, confirmation on the median over two seasons)
8. Trinkwasserverordnung 2023, Anlage 2: values below the limit of quantification are not counted in the PFAS sums (German rule, same logic as French laboratories)
9. US EPA: PFAS national primary drinking water regulation (compliance measured on running annual averages)
10. Li et al. (2018), Occupational & Environmental Medicine: half-lives of PFOS, PFHxS and PFOA after end of exposure (2.7, 3.4 and 5.3 years)
11. EFSA (2020): tolerable weekly intake of 4.4 ng/kg bw for the sum of PFOA, PFOS, PFNA and PFHxS

All sources retrieved September 2026. The full method is on the methodology page. The synthesis was prepared with the help of an AI and then checked against the primary sources.