Finding What Is Driving a High Water Bill: The Register, the Dye Tablet and the Marginal Litre

Ninety-one days, and a number that has never been that shape

The bill is on the counter and it is roughly double. Nothing in the house changed: same people, same showers, same washing machine, nobody filled a pond. The instinct is to start opening cupboards and looking for damp, and that instinct is usually about six steps ahead of the evidence. Before anything gets diagnosed, establish that extra water actually left the main — because a bill can double without a single additional litre moving.

Three billing mechanisms do it routinely. The first is an estimate followed by a correction: a period nobody could read the meter for is billed on an assumption, the next visit finds the true register, and the whole shortfall lands in one document. Look for the read codes printed beside the opening and closing figures, since utilities mark estimated reads distinctly from actual ones, and two estimates followed by an actual is the classic shape.

The second is the length of the period — a quarter is not a fixed thing, and ninety-one days against seventy-eight is seventeen per cent more consumption at exactly the same daily rate. The third is a register that rolled over, was transposed by a digit on entry, or belonged to a meter exchanged mid-period, where the closing read of the old unit and the opening read of the new one have to reconcile and sometimes do not. None of the three has anything to do with the plumbing, and all three are settled by comparing the bill against the meter rather than against last year.

So the first act is arithmetic rather than plumbing. Take the closing read printed on the bill, walk out to the meter, and read it now to every digit the register shows. Divide the difference by the days since that closing read. If the daily figure since the bill was issued sits far below the daily figure the bill charged for, the bill was an estimate catching up and there may be nothing wrong with the house at all. If it is on the same trajectory, water is still leaving at the new rate, and everything below applies.

What one more litre costs, which is not the bill divided by the volume

Every calculation on this page needs a rate, and the obvious way to get one is wrong twice over. Dividing the total bill by the volume folds in the fixed service charge, which is payable whether the house uses anything or not, and inflates the apparent cost of each unit. Then it ignores block pricing, which pushes in the other direction: where a utility charges increasing rates in tiers, the water a leak wastes is not average water — it is the last water, priced in the highest block the household reaches. On a tiered tariff a doubled volume costs rather more than double, and the number that describes the leak is the marginal rate in the top block reached, not the average across the bill.

Then add the sewer side, which is frequently the larger half of the two. Wastewater is very commonly billed volumetrically against the metered water figure, on the reasoning that what comes in mostly goes out: in England and Wales as a stated proportion of metered volume set out in the company's charges scheme, and in North America often as the metered volume itself, sometimes capped for residential accounts at a winter average so summer irrigation is not billed as sewage. The consequence for a leak is blunt. Where that is how the account is charged, every litre through the meter is billed twice, and the marginal rate to use anywhere below is the water block rate plus the sewer volumetric rate added together — with one asymmetry worth knowing before any argument with the utility, since a leak underground or at an outside tap never reaches the sewer at all and is billed for treatment it did not use.

Get the units straight in the same sitting, because the register unit and the billing unit are frequently not the same. North American residential meters commonly register in cubic feet and bill in hundreds of them, the CCF, which is 748 US gallons; some utilities bill per thousand gallons instead, and a metric register counts cubic metres of a thousand litres or 264 US gallons each. A leak rate worked out in litres a minute and priced against a per-thousand-gallon tariff is wrong by a factor nobody catches, because the answer still looks like a plausible amount of money.

The dye tablet, and the two entirely different things a toilet does wrong

Drop a dye tablet or a good few drops of food colouring into the cistern, not the bowl, and then do nothing for twenty minutes. Not ten: a slow seal leak takes longer than the packet suggests, and the commonest false negative in this exercise is an impatient dye test on the toilet that turns out to be the culprit. Colour in the bowl with nobody flushing means water is leaving the tank continuously. While the tablet works, look at what has been living in that tank — an in-tank chlorine block resting against the flapper is the most reliable way there is to destroy one, and both Fluidmaster and Korky say so in their literature, to the point of excluding the damage from warranty.

Now read the water level, because the dye proves flow and the level proves route. A level fallen away from the fill line means water is passing the flush valve seal into the bowl while the fill valve tops the tank up in bursts — the phantom flush, audible at intervals, the fault everyone means by a running toilet. A level sitting at the lip of the overflow tube and staying there means the fill valve is not shutting off and the tank is pouring its inflow down the overflow instead. Both put dyed water in the bowl; they are different repairs, a seal against a seat versus a valve that has lost its shut-off, and the second is normally the larger flow. That tube's height is not arbitrary either — the anti-siphon fill valve's critical level marking has to stand clear of it, which is what ASSE 1002 governs and what stops a tank siphoning back into the supply.

Put a number on it rather than a description, because a running toilet is nothing like a drip and the calculator above wants a flow. Close the isolating stop under the cistern so the fill valve cannot mask the loss, pencil the level on the inside wall, wait ten minutes and mark it again. The volume falls out of one clean identity: a millimetre of drop over a square metre of tank is a litre, or in inches, inside length by width by drop divided by 231 for US gallons. Divide that volume by ten for the rate per minute, which is the unit the leak calculator above is asking for — multiplying a ten-minute figure by six gives litres per hour, and entering an hourly figure in a per-minute field overstates the leak sixtyfold. It only works on a leaking flush valve — where the fault is a fill valve that will not close, isolating the supply stops the leak by definition, and the rate has to come from the meter with that toilet's stop the only thing open in the house.

1. Dye the cistern, not the bowl, and leave it twenty minutes with nobody using that bathroom.
2. Check the bowl for colour, then whether the tank level has fallen from the fill line or is sitting at the overflow lip.
3. Close the isolating stop, mark the level, and time a ten-minute drop to get the seal leak as a volume.
4. For a fill valve that will not shut off, measure that toilet alone at the meter with every other stop in the house closed.
5. Flush and refill before leaving, so nobody meets a dyed bowl and an isolated tank with no note on it.

Where the water goes when every toilet is innocent

If the isolation test says water is moving and the dye tablets say the toilets are clean, the flow is in something automatic, and the one people miss most often is thermal expansion relief. A check valve, backflow assembly or pressure reducing valve on the incoming supply makes the plumbing a closed system; heat the cylinder and the water has nowhere to expand to, so pressure climbs until the temperature and pressure relief valve lifts and dumps to a drain. It reseats, and it does that on every heating cycle for years. The discharge is out of sight by design — a tundish and discharge pipe under the arrangements Approved Document G requires in England, a floor drain under the North American plumbing codes, both of which also require expansion to be controlled rather than merely relieved. A tundish with anything moving through it is a fault, and a valve that has been lifting for months has usually stopped sealing properly as well.

The rest of the list is machinery that runs itself, which is why each item on it gets switched off individually during the hour-long test. Treatment plant is the usual culprit indoors — a softener stuck part way through a cycle, a reverse-osmosis unit whose automatic shut-off has failed, a backwashing filter on a clock — all of which look entirely normal from the front while the drain line runs. Outdoors in a summer bill it is usually a perforated solenoid diaphragm weeping into a zone that drains too well to show a puddle. A pool has a float valve like any cistern, and also loses water that is no leak at all, since an uncovered surface evaporates continuously and the top-up is metered like everything else.

Then there is the buried supply pipe, the expensive answer, which the second isolation test names without ambiguity — internal stop valve closed and the register still moving puts the loss upstream of that valve. Look for the greener strip of lawn, the driveway edge that stays dark in dry weather, clean water standing in a chamber, and listen at the stop valve and the boundary box with a listening stick, since a pressurised leak hisses continuously in a way nothing else in a house does. That pipe generally belongs to the property: in England and Wales it is the customer's supply pipe under the Water Industry Act 1991 and the company's charges scheme, and in North America the tariff says where responsibility divides. It is also the failure most leakage allowance schemes were written for.

One last mechanism inflates everything without being a leak. A pressure reducing valve failed high, or a supply that never had one, pushes more through every fixture and every existing fault at once — orifice flow goes with the square root of pressure, so nothing has to break for the drip to run faster. Both the International Plumbing Code and the Uniform Plumbing Code require reduction above 80 psi, about 550 kPa, and ASSE 1003 is the standard the valve is built to. Put a gauge on an outside tap at the worst hour rather than assuming.

Proving it stopped, and getting the money back

Repeat the isolation test after the repair, same duration, same photographs, because a repair that fixed the louder of two leaks feels like success and reads at the meter as a smaller number that is still not zero. Two faults at once is common in an old house, and the second is invisible while the first is running. Where the utility publishes interval data the confirmation is better than any test: the overnight floor should return to a flat zero on the night of the repair, and that is the graph worth keeping.

Then claim, if there is anything to claim. Most water companies operate a leakage allowance or bill adjustment for a repaired leak, on terms narrow enough to be worth reading before rather than after: usually one claim per property, within a fixed window of the repair, on a plumber's dated invoice, and sometimes conditional on the utility inspecting before the excavation is backfilled. The Consumer Council for Water publishes what the schemes in England and Wales cover; in North America it is the utility's own adjustment policy. The date the leak started, which the interval data gave for free, is frequently the difference between a full allowance and a partial one.

Leave the record behind for whoever meets this next. Write the reads and dates inside the meter box lid, note the marginal rate used and the date it came off the bill, and put the flush volume and the rebuild date inside the cistern lid. Then read the meter monthly for a quarter and write those down too. One reading is a number; twelve is a baseline, and a baseline is what turns the next surprising bill into a five-minute answer rather than another evening with a torch.

1. Re-run the hour-long isolation test after every repair, and keep going until the register genuinely does not move.
2. Confirm the overnight floor returns to zero in the utility's interval data, and save that view.
3. Assemble the dated invoice, the before and after reads and the leak start date, then check the allowance terms before submitting anything.
4. Record the marginal rate used, with the date, so the next comparison is against a known figure rather than a remembered one.
5. Read the meter on the same day each month for a quarter, and keep the list where the bill gets opened.