The position of a measurement point determines which question it can answer. Not how well it measures — which question.
A point at the machine tells you that something is being released, and when. It does not tell you whether it reaches your people. A point where people work tells you that, and not whether it came from your process. For that you need a third point, outside the zone of influence. Three kinds of point, three questions — and most installations answer only one of them.
Why this comes up
In a woodworking plant, exposure is measured on a handful of days a year. The rest of the year, nobody is looking.
That measurement is correct, it is properly carried out, and continuous monitoring does not replace it. What continuous monitoring does is fill in the months in between.
But the moment you decide to fill in those months, one question arrives immediately and it decides everything else: where do the monitors go? In practice that question gets far less attention than the choice of instrument, while it determines much more of what you will eventually be able to do with the data.
What the law actually asks — and what it does not
There is a misconception here worth correcting, because it changes how you look at the interval.
The European carcinogens directive does not set a measurement frequency. Directive 2004/37/EC requires that the nature, degree and duration of workers' exposure be determined so that the risk can be assessed — and then, in Article 3(2), that "the assessment shall be renewed regularly and in any event when any change occurs in the conditions which may affect workers' exposure."
Two triggers, then, not one. Regularly — with no number attached — and on any change in conditions that may affect exposure. A new machine, a relocated workstation, a different board supplier, a modified extraction run: each of those is the second trigger, and each of them is far more common than an annual campaign.
And then the provision that this article is really about. Article 5(5)(e) requires the employer to apply "existing appropriate procedures for the measurement of carcinogens, mutagens or reprotoxic substances, in particular for the early detection of abnormal exposures."
Read that last clause again. The stated purpose of measurement, in the directive's own words, is the early detection of abnormal exposures.
A campaign carried out on a handful of days a year is a sound instrument for establishing a level. It is structurally incapable of early detection of anything, because an abnormal exposure that occurs in the eleven months between campaigns is not detected early — it is not detected at all. That is not a criticism of the campaign. It is what a periodic measurement is.
This is where continuous monitoring belongs in the picture: not as a substitute for the statutory measurement, but as the thing that makes early detection possible at all, and as the data on which you base the interval instead of estimating it.
A note on national implementations. Member States transpose the directive into their own law, and several go further than the minimum — some specify how the interval is to be determined, who must be consulted in setting it, or how it should respond to results near the limit value. Check what applies where your plant stands; the directive is the floor, not the ceiling.
The standard the law points to is called a strategy
Where national law needs a method for deciding how often to measure, it generally points to the same place: EN 689, a European standard adopted across CEN member countries.
Its full title, in the current edition, is worth reading in full:
Workplace exposure — Measurement of exposure by inhalation to chemical agents — Strategy for testing compliance with occupational exposure limit values.
Not a measurement procedure. A strategy. The word sits in the title of the standard the law relies on, which is a reasonable indication that deciding what to measure, where, and how often was never meant to be an afterthought to choosing an instrument.
Three numbers for wood dust and formaldehyde, and the question each one answers
A woodworking plant now deals with three separate figures, and they measure three different things. They are routinely confused.
| Figure | What it is | Which question it answers |
|---|---|---|
| Hardwood dust: 2 mg/m³, eight-hour average, inhalable fraction | Binding EU occupational exposure limit value, Annex III of Directive 2004/37/EC as amended by Directive (EU) 2017/2398. In force since 18 January 2023; a transitional value of 3 mg/m³ applied before that | Are your people inhaling too much wood dust? |
| Formaldehyde: 0.37 mg/m³ (0.3 ppm) eight-hour average, and 0.74 mg/m³ (0.6 ppm) short-term | Binding EU limit value, Directive (EU) 2019/983, with a dermal sensitisation notation | Are your people inhaling too much formaldehyde? |
| Formaldehyde: 0.062 mg/m³ | Something else entirely: an emission limit for the product, under Regulation (EU) 2023/1464, applying since 6 August 2026 to wood-based articles and furniture placed on the market. Determined in a test chamber under standardised conditions | Does the panel you sell or process comply? |
The third is not a stricter version of the second. It does not measure the same quantity, not in the same environment, and not for the same purpose. The first two concern the air your people breathe on the shop floor. The third concerns what an object releases in a test chamber. A workshop comfortably below the exposure limit can be processing panels that fail the product restriction, and the reverse holds too.
Confusing the two — and the figure 0.062 invites it, being six times lower than the exposure limit — leads either to measuring the wrong thing or to drawing the wrong conclusion from a correct measurement.
What you should know about wood dust itself: IARC classifies it in Group 1, carcinogenic to humans, with sufficient evidence for cancer of the nasal cavity, the paranasal sinuses and the nasopharynx (IARC Monographs, Volume 100C, 2012). The evidence is considerably stronger for hardwood than for softwood. That is why the limit value is binding rather than indicative.
And one provision that is regularly overlooked in practice. The directive states it explicitly: where hardwood dust is mixed with other wood dust, the limit value applies to all wood dust present in the mixture. In a workshop that machines oak or beech alongside softwood or panel board — which is to say most workshops — there are not two separate regimes. The 2 mg/m³ applies to the whole.
Where a measurement point sits determines which question it answers
We work with three kinds of point, and each answers one.
At the machine — "is something happening, when, and during which operation?"
This is the point that tells you the dust comes off the sander rather than the saw, and that it happens at a quarter to two. It ties a rise to an action, and therefore to something you can change.
What it does not tell you: whether it reaches anyone.
Where people work — "does it reach them, and with what delay?"
This is the point that distinguishes extraction which captures dust from extraction which merely moves it. At the machine the two are indistinguishable: in both cases the reading drops at the source. At the assembly bench twenty metres away, they are not.
The delay is itself information. Dust appearing at the workstation three minutes after the operation tells a different story from dust arriving twenty minutes later.
Outside the zone of influence — "is this coming from your process, or from outside it?"
This is the reference point. Without it, a peak leaves you unable to say whether to look at your own installation or at the weather.
It is the point most often left out, and it is almost always the cheapest in the setup — it sits in an office or an adjacent room where little happens. And it is the difference between an observation and a diagnosis. A rise visible only inside is yours. A rise the reference point follows with the same amplitude is not.
Why reading temperature and humidity is not a bonus feature
Every measurement point also reads temperature and relative humidity. That is not an accessory function; for formaldehyde it is half the interpretation.
Formaldehyde from chipboard and MDF off-gasses faster in warmth and humidity — the adhesive system in those panels releases more as temperature and relative humidity rise. The consequence is direct: a warm, humid week raises your formaldehyde readings with nothing in your process having changed.
Measure formaldehyde on its own and you see a problem, and you start looking at the extraction. Read heat and humidity alongside it and you see an explanation — and you know to look at the storage area rather than at the installation.
It is the same reasoning as the reference point, one layer down: one parameter gives you an observation, a combination of parameters gives you an explanation.
What to do with this
1. Check which of the three questions your current setup can answer. If monitoring is already running, this is a ten-minute exercise. If everything sits at the machines, you know when something happens — not whether it reaches your people, and not whether it came from you.
2. Put your measurement interval next to your last result, and next to your last change. The directive's second trigger is any change in conditions that may affect exposure. New machine, moved workstation, different supplier, altered extraction: if the interval has not moved with those, it is running on the calendar rather than on the conditions.
3. Add the reference point before you add points at machines. A fourth point at a machine gives you more of the same kind of information. The first point outside the zone of influence gives you a kind of information you did not have.
4. Read temperature and humidity everywhere. It is the cheapest parameter on the list, and for formaldehyde it is the one that separates a problem from an explanation.
InsightAir provides continuous air quality monitoring in critical environments. What happens between two point measurements is usually the most interesting part.
