A conversion is only as good as the number you feed it. Pick the right factor, multiply once, round at the end — and still be wrong, because the 500 g on the display was really 480 g. The arithmetic was never the weak link.
Here is the whole article in one sentence: measuring tools go wrong in three different ways, only one of those ways looks obviously wrong, and a device with more digits on the display fixes none of them. Checking a scale, a tape, or a thermometer takes ten minutes and needs nothing you do not own.
The three ways a measuring tool goes wrong
- An offset. It reads high or low by the same amount every time — 6 g out at 100 g and 6 g out at 1 kg. A bent tape hook, a scale zeroed on a tea towel, a drifted thermometer.
- A scale error. The error grows in proportion to what you measure — 1% out is 1 g at 100 g and 10 g at 1 kg. A stretched or sagging tape, an ageing load cell.
- Poor repeatability. The same thing measured twice gives two answers. A wobbling surface, a dying battery, a worn mechanism.
Only the third announces itself. The first two produce clean, steady, confident numbers that happen to be wrong — which is why they survive for years in a kitchen drawer. Telling them apart is easy: measure at two different sizes. If the gap between reading and truth stays the same, it is an offset. If it grows with the load, it is a scale error.
Accuracy is not resolution — and the display shows you the wrong one
This is the trade-off the packaging hides. Resolution is the smallest step the device can display: 1 g, 0.1 g, 1 mm. Accuracy is how close the reading is to the truth. They are unrelated, and adding digits to a display is far cheaper than making the sensor behind them correct. A scale that reads in 1 g steps can be 5 g out; one that reads in 5 g steps can be spot on. The finer readout does not make it truer — it makes it look truer, which is worse, because false precision is persuasive.
The working rule: treat the last displayed digit as decoration until you have checked the device against something you know. Claiming precision you have not earned is the error covered in the measurement mistakes guide — it just starts at the tool rather than at the arithmetic.
How to check a kitchen scale with a bottle of water
You need a reference mass. A sealed packet with a stated net weight will do at a push, but the printed figure is nominal, with a tolerance around it, and the packaging adds mass of its own. Water is better: its density is close to 1 g/mL near room temperature — about 0.998 g/mL at 20 °C — so 500 mL has a mass of about 499 g. Calling it 500 g puts you within roughly 0.2%, far tighter than the errors you are hunting.
The check itself:
- Put the scale on something hard and level — not a rug, not a draining board, not a worktop join.
- Put an empty container on and press tare.
- Add 500 mL of water. Expect about 500 g.
- Lift it off, put it back, read again. Same number? That is your repeatability test.
- Repeat with 1000 mL.
Now the arithmetic. Say it reads 494 g:
- Error = (494 − 500) ÷ 500 = −0.012 = −1.2%
Then the second load. At a true 1000 g, an offset of −6 g reads about 994 g; a scale error of −1.2% reads about 988 g. One test gives the size of the problem, two tell you which problem it is.
The same error is trivial on a loaf and fatal on a teaspoon
Absolute error stays put while the thing you are weighing shrinks, which is where small errors turn into disasters. Take a scale that is a steady 2 g out.
- On 1000 g of flour: 2 ÷ 1000 = 0.002 = 0.2%. Nobody will ever notice.
- On 10 g of yeast, salt, or baking soda: 2 ÷ 10 = 0.2 = 20%. That is a different recipe.
So the fix for small quantities is not "a scale with more digits" — it is to weigh small things on a scale built for them, or to scale the recipe up and divide. As a rule of thumb, when the quantity is only a few times the smallest step on the display, that display is telling you more about the scale than about the ingredient. Recipes are where this bites hardest, and why recipe conversions keep coming out wrong covers the ingredient-density half of it.
Tape measures: the hook is supposed to wobble
The loose hook on the end of a steel tape is not a fault. It slides by roughly its own thickness so that a measurement hooked over an edge and one butted into a corner give the same answer. Free movement is the design working; a bent hook, loose rivets, or one dropped onto a hard floor is not. Two quick checks:
- Hook it over an edge and measure a fixed span, then butt it into a corner and measure the same span. The readings should match. If they differ, the hook has stopped compensating and you have an offset.
- Compare against a rigid rule, but do not start at the end. Line the 100 mm mark up with the rule's zero and read at the far end, then subtract 100 mm: reading 400 mm means a 300 mm span. That takes the hook out of the question entirely.
Long spans add the second error type — a tape pulled hard stretches slightly, a tape allowed to sag covers a curve rather than a straight line — and both grow with the distance, which is why they never show up on the workbench and always show up across a room. For the units side, the unit conversions guide has the factors.
Thermometers: one fixed point you can make in a mug
The ice point. Fill a mug with crushed ice, top it with cold water, stir, and let it settle a moment. That slush sits at 0 °C, and unlike boiling it barely cares about altitude or weather. Put the probe in the middle of it, not against the mug wall, and read. Anything other than 0 °C is your offset.
Boiling water is the weaker check. Water boils at 100 °C only at standard atmospheric pressure, and the boiling point falls with altitude and shifts with the weather — so unless you know your local boiling point, a reading near 100 °C proves very little.
A thermometer that is 2 °C out, by the way, is not 2 °F out. A temperature difference converts with the factor alone, with no offset:
- 2 °C difference × 9/5 = 3.6 °F difference
Add the 32 and you have converted a difference as if it were a reading — the classic temperature mistake, and a cousin of the look-alike traps behind length, weight, and capacity conversions. Note too that an offset measured at 0 °C proves nothing about the same device at 200 °C; oven thermometers drift most in the range they live in.
What actually degrades, and what only looks worn
Genuinely degrades: load cells that have been overloaded — someone standing on a kitchen scale, once, can deform it permanently, and that is the most common reason a scale that used to be right no longer is. Tape hooks and rivets, after drops. Thermometers after a fall or a spell somewhere far hotter than they are rated for. And batteries, which cause drifting and unrepeatable readings long before a blank screen.
Only looks worn: a scratched platform, faded printing you can still read, paint wear along a tape blade, a cloudy jug. Moulded graduations are a guide rather than a guarantee — but you learn that by weighing water in the jug, not by looking at it.
The routine that helps, and the maintenance that is theatre
Worth doing:
- Keep one known reference. Weigh a sealed bottle of water once, while you trust the scale, and write its mass on the label. A five-second check, forever.
- Re-check after an event, not on a calendar. A drop, an overload, a battery change, a new worktop. Nothing happened? Nothing drifted.
- Check at the load you actually use, and write any steady offset on the device. A measured error is not a broken tool; it is one with a known correction.
Theatre: pressing "calibrate" with nothing to calibrate against — without a reference, the button just moves the error somewhere new. Buying finer resolution to fix accuracy: more digits, same sensor. Taring repeatedly, which removes a container rather than an error. And monthly re-checks of a device nothing has happened to.
Correct it, or replace it?
Correct it when the error is stable and repeatable. A scale that is reliably 1.2% low can simply be divided out: true ≈ reading ÷ 0.988, so 494 ÷ 0.988 = 500 g. Check it back the other way — 500 × 0.988 = 494 — and the correction holds.
Replace it when the tool fails repeatability: the same load, twice, giving two answers. There is no correction for an error that will not sit still. The same goes for an error that keeps changing between checks, or a device you know has been overloaded — in both cases the sensor itself has moved on.
FAQ
How often should I check a kitchen scale? On events, not on a schedule: after a drop, an overload, a battery change, or a move to a different surface. A scale undisturbed in a drawer is not drifting.
Can I calibrate a kitchen scale at home? You can check one with water, and many scales let you re-zero against a known mass. What you cannot do is calibrate without a reference — finding one is the whole job.
Is a scale with 0.1 g resolution more accurate than one with 1 g? Not necessarily. Resolution is the step size on the display; accuracy is closeness to the truth. Different parts of the device set them, and only one is printed on the box.
My thermometer reads 2 °C low in ice water. Can I just add 2? Near 0 °C, yes — that is a measured offset. Do not assume it holds hundreds of degrees away; if the oven range matters to you, check it there too.
Get the number going in right and the rest is arithmetic you can trust. Once you have a reading you believe, convert or scale it in one step with the free calculators at Wapuula Tools — each shows the exact factor and formula behind its result, so you can re-check the working by hand.
