Do rooflights lose more heat than windows?
Yes, and typically by twenty to forty per cent for an otherwise identical unit. A sealed unit that performs at 1.2 W/m²K standing vertical in a wall will perform at somewhere between 1.5 and 1.7 lying near horizontal in a roof. Nothing about the glass has changed. The physics of the cavity, the exposure of the surface and the way heat moves inside the room have all changed.
That gap explains a good deal of confusion in quotations, because the same product legitimately carries two different numbers. This page sets out where the difference comes from and how to make a fair comparison. Target figures are on what U-value should a rooflight have.
A horizontal cavity convects harder
This is the largest single cause. In a vertical sealed unit, the gas in the cavity warms against the inner pane, rises, cools against the outer pane and falls, forming a slow circulating cell. The geometry of a tall thin gap limits how fast that cell can run.
Turn the unit horizontal and the warm gas is now directly underneath the cold pane, which is the classic arrangement for buoyancy driven convection. Instead of one slow circulating cell you get multiple rising and falling plumes across the width of the cavity, moving heat from the lower pane to the upper one considerably faster. The effect is strongest at gap widths above about 14mm, which is why rooflight units are often built with narrower cavities than wall windows.
The pane faces the open sky
A vertical window looks out at the ground, at fences, at other buildings and at a slice of sky. Those surfaces are close to air temperature and radiate back at it.
A rooflight looks at the whole sky dome. On a clear night the effective radiant temperature of a cloudless sky can be fifteen or twenty degrees below the air temperature, so the outer pane radiates heat away and drops below ambient. A colder outer pane means a larger temperature difference across the unit and a faster rate of loss. It is also why frost forms on a rooflight before it forms on a wall window on the same night.
Wind on a roof plane against wind on a wall
Heat transfer at the outer surface depends on how fast air moves across it. Wall windows sit in the boundary layer of the building, sheltered by the wall itself, by returns and reveals, and often by fences and planting.
A roof plane is the most exposed surface on the house. Air accelerates over the ridge and across the slope, and on a flat roof the wind runs unobstructed across the glass. The external surface resistance used in the standards reflects this: the value assumed for a roof is lower than for a wall, meaning the outer surface sheds heat more readily. On an exposed site, and there are plenty across the Essex marshland and out towards Canvey, the real world difference is bigger than the standard figure allows for.
Warm air arrives at the glass first
Inside the room, stratification does the rest. Air warmed by radiators, cooking and occupants rises, so the warmest air in any heated room is at ceiling level. That is exactly where a rooflight is.
A wall window is presented with air at roughly mid room temperature. A rooflight is presented with the warmest air in the house, pressed against it by buoyancy, which increases the internal surface heat transfer and therefore the loss. In a room with a tall ceiling or a lantern the stratification can be two or three degrees, and every one of those degrees is added to the temperature difference the unit is working against.
Frame proportion, roof window against wall window
The frame is generally the worst performing part of any glazed unit, so the proportion of frame to glass matters. Roof windows tend to be smaller than living room windows, and their frames tend to be deeper, because they carry weather protection and a flashing interface that a wall frame does not need.
A 550mm wide roof window sized to fit between trusses can be nearly a third frame by area. A 1.8 metre wall window is a much smaller fraction. That pushes the whole unit figure up on the rooflight even before the cavity and exposure effects are counted, and it is why frame material makes more difference overhead than it does in a wall.
Comparing two figures that were not measured the same way
All of the above is why the plane of measurement has to be stated. A manufacturer quoting the vertical figure for a rooflight is not lying, but the number is not the one the roof will deliver.
Ask which plane the declared U-value applies to. For roof windows the answer should be the pitched or horizontal declaration. For flat rooflights it should be horizontal. If a supplier compares their product with a competitor’s and one figure is vertical, the comparison is meaningless. This single question resolves more specification arguments than any other.
The perimeter around each opening
A window in a wall sits in a reveal with a cavity closer, a lintel over it and a sill under it, and those details are standardised and well understood. A rooflight sits on a kerb standing 150mm proud of the roof deck, or in a trimmed opening between rafters.
The kerb is the harder detail. It stands outside the insulation line and has to have insulation carried up its outer face or it forms a continuous cold bridge round the opening. On a small unit that perimeter loss can rival the loss through the glass. So the honest comparison is not just glass against glass. It is a well understood standard detail against a detail that depends entirely on how carefully it was built, which is why the workmanship carries the ten-year guarantee rather than the glass brand doing the reassuring.
What it means for the whole house
None of this argues against rooflights. It argues for specifying them with the difference in mind.
The area involved is usually much smaller than the wall glazing. A house might have twelve square metres of windows and two of rooflight, so even a forty per cent worse U-value on the rooflight is a small share of the total glazing loss. Against that, a rooflight collects two to three times the daylight of the same area of wall window because it faces the whole sky and nothing shades it, and in a deep plan room that is light no window could have delivered.
Making the comparison fair in practice
Three habits settle it. Compare whole unit figures, never centre pane against whole unit. Compare figures declared in the same plane, and for a rooflight insist on the horizontal or pitched declaration. And add the perimeter to the conversation, because a 1.3 unit on an insulated kerb beats a 1.0 unit on a bare one in any real winter.
Do those three and the rooflight will still lose more heat per square metre than the window downstairs. It will also be the reason the back of the room is usable in November, and the numbers will at least be describing the same thing.
Tell us about your roof
We start with the building, the covering and the slope. Then we tell you what will suit it, and what it will cost, as a fixed written number.
- Surveyed before it is priced
- 10-year workmanship guarantee
- Building Control notification handled
- New installations and replacements