Do roof lanterns lose a lot of heat?

The short answer

Per square metre, yes: a modern roof lantern loses roughly eight to ten times as much heat as the insulated flat roof it replaces, because a good whole unit U-value is around 1.2 to 1.5 W/m2K and a current flat roof build-up is around 0.15 to 0.18. In whole house terms, no: on a typical 1200 by 2000mm lantern in Gloucestershire that difference works out at something in the order of 170 kilowatt hours a year, about what a fridge freezer uses, and a south or west facing lantern gives a fair proportion of it back as free solar gain in the shoulder months. The heat loss people actually notice is not usually the glass at all. It is the cold strip at the top of the kerb, the air leaking through a badly compressed gasket line, and the stratification that puts the warmest air in the house up under the ridge. Those three are workmanship, which is why they carry the ten year guarantee, and they are detailed on the roof lantern installation page.

The comparison that makes the number look alarming

Set the two constructions side by side. A flat roof built to current standards has 150mm or so of rigid PIR above or between the joists and comes in at 0.15 to 0.18 W/m2K. A well specified lantern, counting the frame and the spacer bars as well as the glass, lands between 1.2 and 1.5. The ratio is the number that gets quoted in arguments, and taken alone it is true and misleading.

It is misleading because it is a ratio of two rates over a small area. A lantern typically occupies 2 to 3 square metres of a roof that might be 25 square metres, and the house it sits on has walls, floors, doors and a ventilation rate that all lose heat too. What matters is the absolute figure the lantern adds, not how it compares with the best performing element in the building.

A real size, a real winter, and the answer in kilowatt hours

Work it through on a 1200 by 2000mm lantern. The aperture is 2.4 square metres, and because the glass slopes at 25 degrees the actual glazed area is nearer 2.7. Gloucestershire runs around 2,100 heating degree days in an average year on the usual 15.5 degree base.

At a whole unit U-value of 1.4, the annual heat loss through that lantern is roughly 190 kilowatt hours. The 2.4 square metres of insulated flat roof it replaced would have lost about 22. So the lantern adds in the region of 170 kilowatt hours a year. On a cold January night with 20 degrees inside and zero outside, the instantaneous loss is about 76 watts, which is a bright old fashioned light bulb’s worth of heat leaving through the roof. Those are the honest figures, and they are worth having in front of you before somebody sells a specification on the strength of a decimal place.

Solar gain in the other direction

Glass is the only element of a building that lets heat in as well as out, and on a lantern the sloping planes are well angled for winter sun. A south facing plane at 25 degrees in this part of the country receives somewhere around 1,000 kilowatt hours per square metre a year of solar radiation.

Even with a solar control coating letting only 0.3 of it through, and even counting only the heating season when the gain is useful rather than a nuisance, a south or west facing lantern will collect a couple of hundred kilowatt hours over the winter half of the year. In round terms it offsets the extra loss. A north facing lantern collects far less and is a straightforward net loss, which is one of several reasons why orientation is recorded at survey rather than treated as a detail.

Where the kerb loses more than anybody expects

A 1200 by 2000mm lantern has 6.4 metres of kerb perimeter, and that perimeter is where three constructions meet: the flat roof insulation, the timber kerb, and the aluminium eaves beam. Junctions like that are quantified as a linear thermal transmittance, and a poorly built one runs at several times a well built one.

The difference is not decorative. Where the roof insulation stops at the face of the kerb and the timber runs uninterrupted from a cold outside to a warm inside, that 6.4 metres of timber becomes a continuous cold strip, and it can add as much heat loss again as a meaningful fraction of the glass. Rigid insulation continuous up the external kerb face, lapped tightly to the roof insulation, plus an insulated internal lining and a vapour control layer taped to the eaves beam, is what closes it. The same detail is what stops the winter condensation described under roof lantern condensation in a kitchen.

Perimeter airtightness, and the gasket line at the eaves beam

Conducted heat loss is calculable. Air leakage is not, and it is frequently larger. Air moves through the joint between the eaves beam and the kerb top, through the corners of the frame, and around any actuator cable passing into the kerb.

What controls it is unglamorous: the eaves beam bedded on the kerb top in low modulus silicone and mechanically fixed down, the kerb top planed level so the bedding compresses evenly rather than bridging a dip, corner cleats sealed as the frame is assembled rather than after, and the vapour control layer taped rather than trimmed short. A lantern set on a kerb that dips 5mm in the middle of a two metre run will leak air along that run for as long as it is there, and no glass specification compensates.

Glass builds against what each one loses over a winter

Build up	Whole unit U-value, typical	Annual loss, 2.4m2 aperture	What it costs elsewhere
1990s aluminium lantern, no thermal break, double glazed	Around 3.0	Roughly 400 kWh	Streaming condensation on every internal rafter in January
Thermally broken aluminium, capped, aluminium box spacer	Around 1.8	Roughly 240 kWh	Wider sightlines, and cold spots at every pane edge
Thermally broken, structurally glazed, warm edge spacer, argon	1.2 to 1.5	Roughly 165 to 205 kWh	Nothing. This is the current normal specification
As above with triple glazing	0.9 to 1.1	Roughly 120 to 150 kWh	Around 42 kg per square metre, closer rafters, less daylight
The insulated flat roof it replaced	0.15 to 0.18	Roughly 22 kWh	No daylight at all

Warm air at the ridge, and the stratification a tall lantern creates

Heat loss is not evenly distributed inside the room either. Warm air rises, and a lantern adds 400 to 550mm of height above the ceiling line with glass on every face of it. The air sitting in that volume is the warmest air in the room and it is in contact with the coldest surface, which is an efficient arrangement for losing heat.

Two things reduce it. Air movement at ceiling level, which a ceiling fan run slowly in winter reversal achieves for a few watts, pushes that warm layer back down. And an opening ridge vent, which is exactly the wrong thing to leave ajar in January, precisely because it is sitting where the heat has collected. The vent earns its keep in summer, when the same stratification becomes the free cooling described under roof lantern ridge vents and how they are operated.

Radiant chill, and a room that feels cold at 21 degrees

Comfort is not air temperature. It is roughly the average of air temperature and the mean temperature of the surfaces around you, and a large area of glass overhead is a cold surface radiating away from a person sitting beneath it.

With a whole unit U-value of 1.4 and freezing conditions outside, the internal surface of the glass sits around 14 to 16 degrees while the air in the room is at 21. Sit under it in the evening and you feel that as a slight coolness on the top of your head even though the thermostat is satisfied. A better unit lifts the internal glass surface by two or three degrees, which is a small number that is disproportionately noticeable in a room used for sitting rather than for cooking. It also lifts that surface further above the dew point, which is where the condensation question begins.

Whether a third pane earns its place here

Triple glazing takes the annual loss on that same lantern down by roughly 50 kilowatt hours and raises the internal glass surface temperature usefully. What it asks for in return is real: around 42 kilograms per square metre against 32, which means closer rafter spacing, more frame per square metre, more visible grid overhead and a harder lift onto the roof. It also transmits less daylight.

The honest position is that on a Gloucestershire kitchen extension with a boiler and reasonable use, double glazing with a warm edge spacer, argon and a good thermal break is the right specification, and effort is better spent on the kerb detail and the airtightness. Triple starts to make sense where the room is used quietly for long periods, a study or a bedroom, where a heat pump is running at a low flow temperature and radiant comfort matters more, or where rain noise is also a priority. Current guidance assesses rooflights against a limiting standard measured in the vertical plane, and any current lantern clears it comfortably, so the notification records a number that was never the difficult part. The wider numbers behind all of this sit on the energy efficient rooflights page.