Radiant chill indoors

Why can a room held at the same temperature feel chilly in January and pleasant in June?

The thermostat says twenty degrees in January and twenty degrees in June, and it is not lying — a thermometer beside it agrees. Yet in January the room is the sort of place you put a jumper on in, and in June, at the identical reading, you would call it pleasant. The usual verdict is that the heating is inadequate, or that there must be a draught somewhere.

A neighbouring puzzle — why a tiled floor feels colder than a carpet at the same temperature — is settled elsewhere in this collection, and its answer was about contact. That is not this. Here you touch nothing but a chair and the air, so the question is why a body touching nothing can still be told, by every surface it is not touching, that the room is cold.

Begin with bookkeeping. A person at rest sheds heat by only a few routes: warming the air against the skin, evaporating moisture, conducting into whatever is touched, and radiating. Which does a thermostat measure? Only the first, indirectly — it is a thermometer in air. So if any other route changes between seasons while air temperature is held constant, the thermostat cannot see it, and the occupant certainly can.

Now the route it is blindest to. Every surface radiates according to its temperature, and skin at roughly thirty-three degrees radiates to all of them and receives radiation back. The net is a loss to anything cooler, and indoors at rest it is a share of heat loss comparable to convection. Crucially it ignores the air: it is an exchange between your skin and the walls, floor, ceiling and glass, straight across the intervening space.

So the quantity that matters is not one temperature but two. The second is the mean radiant temperature — the average temperature of the surfaces around you, weighted by how much of your view each fills. What predicts comfort at ordinary air speeds is roughly the average of the two, which is why it is called the operative temperature.

That sentence does the work. In January an external wall's inner face runs below room air, and a window pane, with only a thin insulating layer between you and the weather, runs several degrees below that — and glass fills a good fraction of your view. The fabric has been cold for weeks and has enormous heat capacity, so it is not a surface that briefly dipped but a cold body the heating is losing an argument with. In June those same walls, warmed by mild weather and sun, sit at or above room air. So mean radiant temperature is below air temperature in winter and level or above it in summer: same thermostat, same number, two different rooms.

We can make that quantitative without doubtful figures, because averaging is arithmetic. If comfort follows the mean of the two, then each degree by which the surfaces fall short must be offset by two degrees of air. A room whose surfaces sit three degrees low needs about twenty-three degrees of air to feel like twenty with matched surfaces — a gap paid for in fuel, every hour, all winter.

A second consequence gets misdiagnosed. A cold pane also chills the air against it, which becomes denser and falls across the floor past your ankles. It is felt and reported as a draught, so people hunt for a gap to seal — but the room may be airtight. The window makes it from nothing but its own coldness, which is why radiators were traditionally placed beneath windows: not to warm the glass, but to meet that falling sheet with a rising one.

Is this the mechanism or a story that fits? A clean test exists in old technology: a globe thermometer, a thermometer inside a matt black sphere, which equilibrates under both air and radiation and so approximates the operative temperature. Hang one beside the thermostat. If the account holds it reads below the air thermometer in January, level or above in June, and a curtain drawn across the window moves it while the air thermometer stays put. The refuting observation: if the globe tracks the air thermometer in both seasons and shuttering the glass changes neither reading nor sensation, radiant asymmetry is not what is happening, and the difference lies in air movement, humidity or the occupant.

Three refinements hold the steps together. The first is that "same temperature" was never a complete description of a room: comfort has long been modelled with six variables — air temperature, mean radiant temperature, air speed, humidity, clothing insulation and activity — and a thermostat measures exactly one. Nothing has gone wrong when a room fails at twenty degrees; a single number never specified it.

The second is a rival explanation that deserves naming, because it is real and pushes the wrong way. People acclimatise seasonally, and studies of what occupants accept find that in warm weather they prefer warmer conditions, not cooler; clothing points the same way, since you wear less in June. Both predict that twenty degrees should feel worse in June. It feels better. That the two most obvious rivals have the opposite sign is the strongest evidence that something in the fabric, not the person, is doing the work.

The third is a caveat about magnitude. How much of the January deficit comes from cold surfaces, how much from the downdraught, and how much from winter infiltration depends entirely on the building — single glazing in a solid wall behaves nothing like a modern airtight house, where the effect nearly vanishes. The mechanism is general; its size is not.

• Why radiant asymmetry causes discomfort out of proportion to its effect on the average.

Every term the collection defines is gathered in the glossary.