Why the night is worth more than the day
Every daytime reading you take is contaminated. Someone runs a tap, the heating fires for twenty minutes, a washing machine fills, the cylinder reheats. All of that moves pressure around, and none of it tells you whether water is leaving the system.
A night removes those variables for free. Nobody draws water. The building is still. The temperature of the pipework settles and stays settled. If the number is lower in the morning than it was last thing, and nothing ran in between, the water went somewhere it was not supposed to go.
That is the whole value of the test, and it is why it is worth doing properly rather than approximately. A sloppy overnight test gives you a number you cannot trust, and an untrustworthy number is what sends people to lift the wrong floor.
The standing test, step by step
This is the sealed heating side first, because that is the gauge most people are looking at. The mains-fed side is a different test and it is further down.
- Let the system go properly cold. At least two hours after the last time the boiler fired, and ideally longer. Water contracts as it cools, so a reading taken on a warm system will fall on its own overnight and tell you nothing.
- Turn the programmer off entirely. Not down — off. A hot water schedule at five in the morning, or a frost thermostat kicking in on a cold night, makes the system hot again and hands you a false result. If you cannot switch it off, note the times it ran and treat the night as spoiled.
- Read the gauge and write it down with the time. Photograph the dial straight on if the needle sits between markings. Your phone records the time for you, which is half the point.
- Change nothing else. Do not bleed a radiator. Do not open or close valves. Do not touch the filling loop. The test measures a system you have left alone.
- Do not top up, whatever the reading says. Topping up erases the measurement and puts fresh oxygenated water into the pipework. Only refill if the boiler has locked out and you genuinely need heat, and if you do, the night is spoiled and you start again.
- Read it again eight to twelve hours later, before anything runs. Same dial, same angle, same light, and write down the number and the time again.
- Do it three nights running. One night is an anecdote. Three nights is a rate, and a rate is what tells an engineer where to look.
Two readings and two times per night, on paper or in a note on your phone. That is the deliverable. Bring it to whoever attends and you have saved yourself an hour of someone standing in your hall watching a needle.
Reading what you measured
| What three nights showed | What it narrows to |
|---|
| No movement you can see, three nights running | Nothing is leaking while the system is static. Any loss you have is happening when the system is hot, which points at the expansion vessel or the relief valve rather than at pipework. |
| A small fall, the same size each night | A genuine leak, running at a steady rate. A repeatable number is the most useful result this test produces. |
| A large fall every night, or it will not hold at all | An active leak. Stop running the circuit and get it isolated before more water goes into the fabric of the building. |
| A fall on one night out of three | Usually the programmer ran, or the first reading was taken on a system that had not finished cooling. Repeat before drawing any conclusion. |
Worcester Bosch describe needing to top a system up once or twice a year as ordinary. That is a different thing from a fall you can see by morning, and the distinction is the one worth holding on to.
A measured loss with nothing visible is not a dead end
This is the point most people give up at. The gauge says water is leaving. Every radiator valve is dry, the boiler case is dry, there is no stain on any ceiling. It feels like a contradiction and it is not.
What a measured loss with nothing visible actually rules out is the easy half of the search. Surface leaks announce themselves — they drip, they stain, they smell. A loss that hides is a loss into something absorbent or something drained: a screed floor that soaks and then dries under the heating, a joint inside a wall chase, a pipe running under a suspended floor where the water drips onto soil or into a void and never reaches a surface you can see.
So the finding is genuinely informative. You have a loss rate, and you have eliminated everywhere you can look at. What you are left with is the concealed pipework, which is exactly the search that thermal imaging, acoustic listening and tracer gas exist to do. The overnight drop page covers what the timing rules in and out on the heating side; this page is about making the measurement good enough to act on.
The mains side, which has no gauge
If the system you are worried about is the cold mains and a stored hot water cylinder rather than a sealed heating circuit, there is usually no dial to read. The test changes shape but the logic is identical.
Use the meter instead of a gauge
Last thing at night, with every tap off and no appliance due to run, read your water meter and photograph it. Read it again in the morning before anyone gets up. Any movement across a still night is water leaving the private pipework you are responsible for. That is the same test as the boiler one, with a different instrument. More on that reading at a meter turning with everything off.
Where it goes on an unvented cylinder
An unvented cylinder sits at mains pressure permanently, so anything marginal on it is under load all night. The two places to look at first are outside, at the discharge pipe, and inside, at the tundish under the safety valves. A discharge that drips overnight is the system relieving pressure it cannot otherwise absorb, and that is a loss you can see if you go and look.
What you should not do is start adjusting anything on a cylinder. Work on unvented hot water storage is controlled under Part G of the Building Regulations and belongs to someone qualified to do it. We say the same about the boiler case: that is Gas Safe territory, not ours.
Why failures cluster overnight in the first place
There is a second reason the night matters, and it is not about your house at all. Mains pressure in a street is at its highest when demand across the network is at its lowest, because less water moving through the mains means less friction loss between the pumping station and your stopcock. The quiet hours are when the static pressure standing against every joint, every washer and every tired compression fitting in the building is at its peak.
So a marginal fitting gets its hardest test at three in the morning, with nobody awake to hear it. That is why a burst is so often discovered in the morning rather than caused in the morning, and why a pipe that has behaved all winter can let go on a quiet night. It is also why the overnight standing test is fair: you are testing the system at the pressure it actually has to survive.
Things that quietly ruin the test
- Topping up before reading. The commonest one. You cannot measure a loss you have already refilled.
- A filling loop left slightly open. It feeds the leak and flattens the reading. Check both valves are properly shut and, if yours detaches, detach it.
- Starting warm. Contraction as the system cools looks exactly like a leak on the first night.
- An ice-cold night with the heating off entirely. A hard frost pulls the whole system temperature down further than a mild night does, which moves the needle slightly on its own. Three nights averages that out.
- Reading a different gauge. A digital display and an analogue dial on the same boiler do not always agree. Pick one and stay with it.
What happens when you call us with three nights of readings
The readings tell us whether to bring detection kit at all, which is the most useful thing a customer can hand over. If the loss is real, the circuit is isolated and pressure-tested so your nightly figures become a measured loss rate under controlled conditions. Thermal imaging follows pipe runs through screed and plasterboard, acoustic equipment listens for escaping water on the mains side, and tracer gas goes into a drained circuit where nothing else will find a pinhole under a poured floor. Then the floor is opened once, in one place, and our own handymen put the screed, boards and finish back.
£150 an hour, agreed before we attend, no call-out fee, and if we do not find the leak you do not pay for the looking time. The visit includes a written report set out to support a trace-and-access claim, though what your policy covers is your insurer's decision and not ours. Rates in full on the pricing page, and the repair side at boiler and heating leak repair.
