Where this starts
A window is two things: an opening part that has to seal against its frame, and a frame that has to seal against a hole cut in a wall. Both leak, they leak in different places, and the second is invisible.
SMOKE IS THE TEST. On a windy day, or with an extractor fan running to depressurise the house, run a smoke pencil, an incense stick or a thin strip of tissue slowly around the whole window — along the opening joint, around the frame where it meets the plaster, into the corners of the reveal, and along the cill. The air moves where the leak is, and it is very often at the frame-to-wall junction rather than at the sash.
WHY THE JUNCTION LEAKS. A window is fitted into an opening slightly larger than itself, and that gap is meant to be sealed and insulated before the reveal is made good. It frequently is not: the gap is stuffed loosely, or filled with expanding foam that has been trimmed and left unsealed, and the plaster and architrave hide the result. Air then travels from the cavity or the outside, along that gap, and into the room at whichever edge is least sealed.
TRICKLE VENTS ARE NOT DRAUGHTS, and this is worth settling before anything is sealed. A background ventilator is a deliberate opening providing the room's air exchange, and taping it over transfers the moisture that was leaving through it onto the coldest surface in the room. A vent that feels uncomfortable in cold weather has a flap, and closing the flap temporarily is what the flap is for.
COLD IS NOT THE SAME AS DRAUGHTY. Air falling off a cold pane produces a convection current down the glass and across the floor that feels exactly like a draught and involves no air entering the building at all. The distinction is that a real draught has a source, moves smoke horizontally at a specific point, and changes with the wind. A cold downdraught is present whenever the glass is cold, everywhere along the window, and does not care about the wind.
What it could be, and what tells them apart
These are candidates to check against your own building, in the order they most often turn out to be the answer. None of them is a diagnosis of your situation.
The gap between frame and wall was never sealed
Most commonSmoke moving at the junction between frame and plaster, or out of a reveal corner, rather than at the opening joint — often on a replacement window fitted into an existing opening.
How to check: Run smoke around the frame perimeter, not the sash. Movement there is a gap behind the plaster.
Perished or compressed weather seals
Most commonSmoke moving at the joint between the opening part and the frame, with the gasket flattened, cracked, shrunken at the corners, or missing altogether.
How to check: Close the window on a strip of paper and pull. If it slides out easily at some points and not others, the seal is not compressing evenly.
A cold downdraught mistaken for a leak
CommonA chilly current felt at floor level near the window with no smoke movement anywhere, present in still weather as much as in wind, on single glazing or a failed unit.
How to check: Whether smoke actually moves. Cold air falling off the glass feels identical and is not air entering the building.
Hinges or locks out of adjustment
CommonA leak at one edge only, often the handle side or one corner, with the opening part not pulling evenly onto its seal when locked.
How to check: Whether the paper test fails at one edge only. A sash that is not pulling in evenly is adjustment, not seals.
A trickle vent left open, or the vent being blamed
CommonAir entering at a slot along the top of the frame, with a flap that opens and closes, in a house that may otherwise have no background ventilation at all.
How to check: Whether the opening has a flap. If it does, it is a designed ventilator and taping it shut moves the moisture problem elsewhere.
A leak elsewhere presenting at the window
Less commonAir arriving at the window reveal but traceable further back — an unsealed cavity at the lintel, a service penetration, or a loft hatch pulling air through the wall.
How to check: Follow the smoke backwards. Air arriving at a window has sometimes travelled a long way to get there.
What to check, in order
The order matters — each step is here because of what it rules out for the ones after it.
- 1
It is the only way to tell a sash leak from a frame-to-wall gap, and the second is both the commoner and the one no amount of work on the window will fix.
- 2
Closing the window on a strip of paper and pulling tells you where the seal is compressing and where it is not, which separates worn gaskets from an out-of-adjustment sash.
- 3
Cold air falling off glass feels exactly like a draught, involves no air entering the building, and is fixed by glazing or a curtain rather than by sealing anything.
- 4
A background ventilator is providing the room's air exchange, and taping it over relocates the moisture onto the coldest surface in the room.
- 5
Depressurising the house exaggerates every leak and makes a marginal one obvious, which is the cheap domestic version of a blower door test.
The calculations behind this
Several of the causes above are a quantity that was specified or assumed. These compute them, with the formula and the standard behind each.
When to stop and get somebody in
- Where sealing the window would mean removing the architrave and making good the reveal, since that is a plastering and decorating job as well as a sealing one.
- Where the window is a replacement and the frame-to-wall gap was never sealed, which may be a workmanship claim against whoever fitted it rather than a maintenance item.
- Where the house has no background ventilation at all and the draughts are effectively providing it, because sealing them without adding ventilators produces condensation.
- Where there is damp or staining at the reveal as well as a draught, which means water is following the same path the air is.
This page lists candidates to check, not a diagnosis — nobody has seen your building. If the symptom involves anything structural, read this first.