Do roof lanterns leak?

The short answer

A correctly detailed roof lantern does not let water in, and where water does appear it is almost never through the glass or the lantern frame. Three things account for nearly every case: the kerb and the junction between it and the flat roof covering, standing water sitting against that kerb because the roof does not fall away from it, and condensation being manufactured inside the room and landing on the coldest surface, which is the aluminium overhead. The first two are workmanship and they are exactly what the ten year guarantee is written against. The third is not water getting in at all. Distinguishing between them is straightforward once you know what to look at, and the underlying detailing is set out on the roof lantern installation page.

Three places water actually comes from, and how to tell them apart

Start with when it appears. Water that only shows during and after rain, and gets worse when the wind is from one particular direction, is coming from outside. Water that appears on cold mornings, is worst in January, and dries up by lunchtime is condensation. Water that shows up days after the rain stopped, or in a dry spell, is usually something that has been sitting inside the construction and is finding its way out slowly.

Then look at where. A mark at the internal top edge of the kerb lining, running all the way round, is a cold bridge. A mark in one corner only is a detail at that corner. Water running down the inside face of a glazing bar is drainage inside the frame that is not getting away.

A lantern is drained, not sealed: where the water inside the frame goes

This is the part that surprises people. A modern lantern is a drained and ventilated system, and it is designed on the assumption that some water gets past the outer gasket. It always does, driven by wind pressure across a joint that has to accommodate movement.

What happens next is the design. Each rafter carries a drainage channel below the glass line. Water that gets past the outer seal lands in that channel, runs down the fall of the rafter, discharges into a collector at the eaves beam, and leaves through weep slots on the outer face above the kerb top. It never reaches the room. A lantern that drips internally usually has that path blocked, most often by silicone applied over the weeps by somebody trying to stop a mark, or by debris in the eaves collector. Sealing the escape route is how a small amount of designed drainage becomes water on the worktop.

Reading the mark: position, timing and weather

What you see	When	Most likely cause	What it means for the work
Damp band all round the internal top edge of the kerb lining	Cold weather, worst in January	Cold bridge where insulation stops at the kerb face	Kerb lining opened, insulation made continuous, vapour layer taped to the eaves beam
Droplets on the room side of the glass, gone by midday	Cold clear mornings	Condensation from cooking, showers and drying clothes	Ventilation and extract, not a defect in the lantern
Water running down a glazing bar during rain	Heavy rain, any direction	Blocked drainage channel or weep slots	Channels and weeps cleared, gasket line checked for compression
Water at one corner of the kerb only	After prolonged rain	Corner detail in the waterproofing, folded rather than formed	Covering taken back at that corner and the junction re-formed
Water only when wind drives rain from the south west	Storms	Upstand height below 150mm at the high point of the fall	Kerb rebuilt to height, covering re-dressed over the top
Wet mark on the ceiling a metre away from the lantern	Any time, delayed	Water tracking along a joist from a failure elsewhere in the roof	The lantern is not the source. The flat roof gets investigated first
Cloudiness between the panes that never wipes off	Permanent, gets worse	Failed edge seal on the sealed unit	New glass unit into the existing frame

Ponding against the kerb, and why the fall matters more than the height

A 150mm upstand is only 150mm if the water starts at the roof surface. Where the roof ponds, the water level rises, and 20mm of standing water against a kerb turns a 150mm upstand into a 130mm one that spends six months of the year wet.

Standing water is patient. It finds the smallest weakness in a lap, it freezes and expands in it, and it thaws into the construction. A flat roof is designed to fall at 1:40 so that after deflection and normal timber movement the finished fall is still no shallower than 1:80. Where an existing roof already ponds at the position a customer wants a lantern, the honest answer is that the deck needs re-falling before the kerb goes on, and that is said at survey. The specific case of a roof with no fall at all is covered on can a roof lantern go on a flat roof with no fall.

The eaves beam bedding, and the joint that is doing the sealing

The base of the lantern is bedded onto the kerb top on a continuous bead of low modulus silicone and mechanically fixed down through it. That bedding is a seal between two rigid things, and it works because the kerb top is flat, level and waterproofed underneath it.

Where it fails, the cause is almost always upstream. A kerb that dips 5mm in a two metre run leaves the base frame bridging a hollow, the bedding never compresses there, and the gasket line above it sits unevenly for the life of the lantern. Packing under a base ring is not a solution, it is a deferred problem. The kerb is planed or shimmed and re-checked with a straight edge and a level before the frame is unwrapped, which is why the kerb goes up first and the glass is ordered after.

Condensation on the inside of the glass is not a defect

Warm air holds moisture. A kitchen produces a great deal of it, from a hob, a kettle, a dishwasher and clothes on an airer, and that air rises to the highest point in the room. Under a lantern the highest point is a sloping glass surface with cold sky on the other side of it, so that is where the moisture lands first. On a cold clear morning a lantern can run with water while every other surface in the room is dry, and that is physics behaving correctly.

Three things control it: extract at source over the hob, ducted outside rather than recirculating; background ventilation, either through a vent in the lantern or trickle vents in the room; and getting the insulation continuity right so the aluminium is the only cold surface and not the plasterboard as well. Persistent black spotting at the top corners of a kerb lining is a construction problem, not a lifestyle one.

Thermal movement, and why a bead of silicone is never the waterproofing

A 3 metre aluminium ridge sitting in July sun can be 40 degrees hotter than it is on a January night, and aluminium expands by roughly 2.4mm over that length across that range. Every joint in a lantern has to absorb that movement, which is why they are gasketed and drained rather than glued.

Silicone applied externally over a joint that is designed to move will split within a couple of years, and while it is intact it blocks the drainage that was doing the real work. The same applies at the kerb: the waterproofing is dressed up the face and over the top so the upstand is part of the roof, and it is never a sealant line at the junction between two materials. Where a bead of silicone appears to be holding a lantern dry, the detail underneath it has already gone.

What the ten year guarantee is actually written against

Four details decide whether a lantern is still dry in twenty years, and all four are buried by the time anyone sees the finished work: the kerb height measured from the finished surface at the high point of the fall, the falls in the surrounding roof, how the covering terminates over the kerb top and at the corners, and insulation continuity around the kerb with the vapour layer taped to the frame. None can be inspected once the plaster is on, which is exactly why they get skipped by people in a hurry.

The workmanship guarantee runs for ten years against those four, because they are the ones workmanship controls. The glass and the frame carry the manufacturer’s own warranty, which is a separate document with its own terms, and the survey notes say which is which.

Severn Vale flood history, and the scepticism it earns

Homeowners in Gloucester, Longford, Tewkesbury, Sandhurst and the vale villages watched water come into their houses, most memorably in 2007, and a good number of them are wary of anybody breezy about cutting a hole in a roof. That wariness is entirely reasonable and it is a better starting point than enthusiasm.

The answer to it is not reassurance. It is the perimeter detail written down before the work starts, an opening that is never started unless it can be closed and made watertight the same day, a lantern that is water tested rather than assumed once the glass is in, and a Building Control inspection regime with a completion certificate at the end of it. Where a survey finds an existing deck that is already wet, that goes in writing too, because a lantern on a roof that is failing underneath it is a lantern that will be blamed for something it did not do.