A drip lands on a reception desk. Someone looks up, sees the glass roof, points at the nearest joint and books a contractor to seal it. The sealant goes on, everyone relaxes, and four weeks later the same drip is back in the same spot.

This happens constantly, and it isn’t incompetence. It’s that water inside a building envelope behaves nothing like water in a pipe.

Water travels sideways, and it travels a long way

Once water gets past the outer line of defence it’s inside a system of frames, channels, insulation voids and structural members. Those run horizontally. Water follows them, held against the underside of surfaces by surface tension, until something interrupts it: a bolt head, a bracket, a gap in a seal, a change in material.

That interruption is where it drops. It can be several metres from where it got in, on the other side of a structural bay, and on a large atrium roof the distance can be most of the span.

Height makes the displacement worse. A tall atrium gives water a long way to run inside the framework before it finds an exit, which is why the drip in the middle of the lobby is often fed by a defect over by the perimeter.

The defect is usually tiny and usually boring

People expect a dramatic cause. Cracked pane, hole in the roof. Those are the rare ones, and they’re the easy ones, because you can see them.

The common causes are undramatic. Perimeter seals that have gone hard and pulled back a few millimetres at a corner. Gaskets that have shrunk, which they do, so the compression they relied on isn’t there any more. A blocked outlet backing water up until it rises above a weathering line that was never meant to be submerged. Failed edge seals on glazing units. Movement at the junction between the glazed roof and the parapet or wall it meets, where two materials expand at different rates and the joint between them has been working back and forth for fifteen years.

Capillary action gets its own mention because it defeats common sense. Water will travel upward through a narrow enough gap, which means a defect can sit above the level where water is pooling and still take it in.

How a competent trace actually works

The method is elimination, done in order and done patiently, and it’s the opposite of sealing the nearest joint.

It starts dry, from inside. Where exactly does the water appear, and does it coincide with a bracket, a bolt, a bay junction? What’s the pattern on the ceiling or wall, and how old is the staining compared with the current leak? An old stain with a new drip is a different story from fresh damage.

Then the suspect area gets isolated and water goes on deliberately. Hose testing works from the lowest point upward, section by section, with someone inside watching while someone outside applies water to one zone at a time. The sequence matters. Start high and you’ve wetted everything below, and the test tells you nothing.

Each zone gets long enough to be meaningful, because a slow path can take twenty minutes or more to show up inside. Rushing the dwell time produces a clean result on a defective joint, which is the worst possible outcome.

Dye tracing helps where several possible entry points sit close together. Different colours in different zones, and the colour that arrives inside identifies the route.

Thermal imaging has a real use and an overstated one. It’s good at showing where moisture has accumulated in a build-up, so it narrows the search area. It doesn’t show you a 2mm gap in a seal.

The structural context sits underneath all of it. Understanding how glazed assemblies are put together and where the load paths run tells you which cavities water can get into, and the same logic that decides where glass belongs against a solid wall in a commercial interior applies overhead, except that gravity is now working against the joint rather than along it.

Why the shortcut keeps winning

Tracing properly takes time, and time at height over an occupied space costs money. A tube of sealant and an hour’s work is cheap, so it gets tried first on a lot of buildings.

The trouble is that each failed attempt makes the next trace harder. Sealant applied over a drainage path blocks it. Water that used to escape through a designed route now sits in the framework instead and finds a new way out somewhere else. After three or four rounds of that, the original defect is buried under remedial work and the leak has moved.

Contractors who do this properly, including high-level specialists like Sky Access, will quote the investigation separately from the repair, and that’s the right way round. Access planning runs alongside it, since much of this work happens over live floors and falls under the Work at Height Regulations.

If a contractor arrives at a leaking atrium and goes straight for the sealant gun without asking where the water appears inside, they have already decided what the problem is. They haven’t looked.