Climate monitoring for collections

Everything else in preventive conservation depends on knowing what the environment actually does. Not what the building manager says it does, not what the wall-mounted dial in the corner says, but what a calibrated instrument records over a full cycle of seasons. This section covers how to get numbers you can defend.

The four things that make data trustworthy

A climate record is only as good as its weakest link, and there are four of them.

Accuracy. Capacitive humidity sensors, which is nearly all of them, typically claim something in the region of ±2 to ±3 percent RH under laboratory conditions. Real accuracy in a cold store or a damp cellar is usually worse. A sensor that reads 4 percent high will make a compliant room look non-compliant, and someone will spend money because of it.

Drift. Humidity sensors do not hold their calibration indefinitely. Drift of around one percent per year is common, more in polluted or very humid environments. Without a calibration routine you will eventually be logging the sensor’s decline rather than the room’s climate.

Placement. A sensor on an external wall reads the wall. A sensor above a radiator reads the radiator. A sensor inside a case reads the case, which is often exactly what you want but not the same as the room. Where you put it changes the answer more than which model you bought.

Interval. Logging every four hours will miss the sharp morning spike when the heating comes on, and it is precisely those short excursions that stress hygroscopic materials. Fifteen to thirty minutes is a reasonable default for most spaces.

Choosing between the instrument types

There are broadly three ways to record climate, and they suit different situations.

Standalone data loggers are self-contained: they sit in place, record to internal memory, and get read out with a cable or a short-range wireless handshake. They are cheap per point, they do not depend on the building’s network, and they keep working during a power cut. The cost is labour, because someone has to walk round and download them. This is the right starting point for most small collections, and the loggers page goes into the selection criteria.

Wireless sensor networks push readings to a base station or a cloud service continuously. The advantage is not really convenience, it is alarms: you find out about a failed dehumidifier the same day rather than at the next download. The cost is infrastructure and, usually, a subscription. Wireless networks covers what to check before committing.

Handheld thermohygrometers measure a point in time. They cannot substitute for logging, but they are indispensable for checking a logger, surveying a building quickly, or answering a question during an installation. Thermohygrometers covers what separates a usable instrument from a toy.

Absolute humidity is the quiet key

Relative humidity is the number everyone quotes, and it is genuinely the one that matters for most materials, because it governs how much moisture an object exchanges with the air. But it is a ratio, and the ratio moves when temperature moves even if no moisture enters or leaves the room.

This has a practical consequence that catches people out constantly. If a room is at 20 °C and 50 percent RH, and you drop the temperature to 15 °C without changing the amount of water in the air, the relative humidity rises to roughly 69 percent. Nothing got wetter. The air simply got better at holding what it already had.

The reason this matters is diagnostic. When you see relative humidity swinging, the first question is whether the moisture load changed or the temperature did, because the fixes are entirely different. Plotting temperature and RH on the same axis, and ideally calculating dew point or mixing ratio alongside, turns an unreadable sawtooth into an obvious story. Reading climate data works through this.

A monitoring plan that survives contact with reality

Ambition is the enemy here. A plan with forty sensors that nobody downloads produces nothing; a plan with four sensors that get read every month produces a defensible record.

• Start with one logger per climate zone plus one outdoors. Add sensors only when the data raises a question you cannot answer.
• Write down where each sensor is, to the shelf. In three years nobody will remember, and a record with unknown provenance is nearly worthless.
• Fix a download day. Monthly is enough for most collections; quarterly is the absolute minimum with typical memory depths.
• Check one logger against a reference instrument at every download, and put the whole set through a proper calibration check annually.
• Keep the raw files. Exported summary graphs age badly, raw CSVs do not.

Where to go next

Once you have a year of data, the decisions start. Target humidity ranges covers what you should be aiming for and why the classic figure has been revised. If the data shows a problem, humidity control covers the equipment side. If it shows the building is fine but individual cases are not, showcase microclimates is the relevant page.