Same water, one night or one winter

You wake up and the bedroom window is streaming, there is a puddle on the sill, and when you pull the mattress away from the outside wall to check, the back of it is damp. None of that happened while you were watching. It happened between about one and five in the morning, in a rig that felt perfectly comfortable when you went to bed.

That water and the mold that turns up behind the same mattress in February are the same water. Condensation is it arriving in a night. Mold is it arriving over a winter, in the coldest and stillest corner of the rig, somewhere nobody looks until something smells wrong.

Both are decided by the same physics, and it is unusually simple physics: the air in the rig has a dew point set by its temperature and its humidity, and every surface colder than that gets wet. Which means both are predictable, and one of them is predictable early enough to do something about.

Tonight: will the windows sweat?

a. One sensor at the glass, and no arithmetic

The easiest useful thing on this page: put a cheap thermo-hygrometer on the window sill or against the outside wall, and warn above 80 per cent. Air a couple of inches from a cold surface climbs toward saturation as that surface approaches the dew point, so 80 is the early call — it trips while you are still awake, and it needs no calculation at all.

It reads the air near the glass rather than the glass itself, single and double glazing behave differently, and cheap sensors are least accurate at the very top of the humidity range. There is also a quirk worth knowing: many of them read wrong for hours after they have been properly wetted, so a sensor stuck at 95 to 100 per cent long after the room has dried is not broken so much as reporting, badly, that the glass was streaming right there.

b. The dew point, which you work out yourself

Nothing in the catalog broadcasts a dew point and the app does not calculate one, so this is a number you derive from a sensor you already own. A bedroom at 68 to 70°F sitting at 60 to 65 per cent humidity has a dew point of roughly 54 to 58°F. Anything in the rig colder than that will be wet by morning.

Once you have the figure, everything else on this page is a comparison against it. The catch is that it moves: run a shower and it climbs, open a vent and it falls, and the app will not re-derive it for you. Work it out once for how your rig sits on a typical winter evening and use that rather than chasing it nightly — and if you would rather the app did the arithmetic, that is what the request board is for.

c. A probe taped to the glass, for the honest answer

A wired probe against the surface stops you inferring. A Govee H5044 reads every 30 seconds, the fastest here, and an Inkbird IBS-P01B every 120. Set its low threshold at the dew point you worked out: below that, condensation is forming right now whatever the glass looks like from the bed, and above it the glass is dry whatever the humidity number says.

d. An outside sensor, as a season-long proxy

A single-pane window’s inner surface sits somewhere between the outdoor and indoor temperature, and closer to outdoor than anyone would like. So an outdoor sensor warning below your dew point — below 55°F, for the bedroom above — is a decent standing rule. A RuuviTag is sealed and belongs outside; a Govee puck needs shelter.

It is honestly a proxy rather than a measurement. Wind and a clear sky make glass colder than the air around it, so the margin is generous on a still night and thin on a windy one. Set it for the season and leave it.

This winter: where mold actually grows

Mold is not about a number spiking. It is about humidity staying above roughly 60 per cent at a surface for days rather than minutes, and it happens in the coldest, stillest spot in the rig — behind the mattress, the back of a closet, a basement bay, behind the fridge. Places with no airflow and an outside wall.

Which is why the sensor in the living area is the wrong sensor for this. It will read a comfortable 50 per cent for the entire winter and be perfectly accurate, right up until you find the soft spot. The cheap multipack is the right purchase here: warn above 60 per cent in the living area, and a little higher — 65 — in the closet and the bay, because those places sit higher by nature.

Set the warning where a shower does not trip it and a wet week does. Spikes are cooking, towels, a rainy day with the door open. The reading to act on is the slow climb that never comes back down. And accept that a cheap sensor is worth about plus or minus five points on humidity — enough for this job, not enough to argue with, and the Jinou‘s humidity in particular is read from a stable byte position rather than checked against a reference, so treat it as indicative.

The combinations

Humidity on its own says the air is damp. What makes it actionable is the second reading that says why — the pattern the cookbook runs on.

• Closet or bay above 65 while the living area sits in the 50s — a local cold spot, not the rig’s air. Fix airflow in that spot: a gap behind the mattress, a bay vent, a louvered door. A bigger dehumidifier in the living area will never reach it.
• Everything above 60, and the bedroom CO2 climbing overnight — ventilation. The rig is sealed and breath is the source, so crack a vent. The CO2 article is the longer version.
• Everything above 60 and CO2 normal — there is a water source rather than a breathing problem. A wet bay, a leak, towels drying, or an unvented heater.
• Humidity climbing overnight while the propane falls faster than the cold explains — the catalytic or portable propane heater is the source. It makes water as it burns. Vent it, or use the furnace, which exhausts outside.
• Humidity climbing over days and the leak puck trips — you have a plumbing problem, not an air problem. Fix the leak and watch the humidity fall over the following days as confirmation.
• Bedroom humidity rising since bedtime and the outdoor sensor dropping below your dew point — wet windows by morning. Crack the roof vent now, run the furnace fan, put the reflective panel up. All three change tonight’s outcome.
• A bay sensor was reading high and then went quiet — the cell died in the cold. The bay’s condition is now unknown rather than fine, and silence is not a good reading.

What none of this tells you

Nothing here sees mold, or water inside a wall cavity, or moisture under the floor. Humidity in the air is not moisture in a surface, and the condensation that costs a floor happens where no sensor can go. These readings tell you the conditions that grow it, which is genuinely useful and is not the same as seeing it.

What to do about it

Buy the cheap multipack and put the sensors where mold actually grows rather than where you sit. That is the whole of the mold half, and the sensor you already own in the living area does not do it.

For tonight’s windows, the window-side sensor at 80 per cent is one number and no arithmetic, and it is the one that trips early enough to matter. Add the surface probe only if you want to know rather than infer. And a leak puck under the wet bay stays worth having, because it is what separates “the air is damp” from “there is water coming in”, and those are different jobs on different Saturdays.

Then do the thing the readings are actually for: open a vent before bed on a cold clear night. Every method on this page exists to tell you which nights those are, and none of them does anything about the water on its own.

Where to get one

Quantity beats quality here — four cheap sensors in the right corners tell you more than one good one in the wrong place. The exception is anything living outside, where sealed is worth paying for. The compatibility database lists every sensor that reports humidity; picking between them is a separate argument.