Science skills

Uncertainty and percentage errorHow to work them out, step by step, and what they tell you

Every measurement has some doubt in it. Uncertainty puts a number on that doubt. Percentage error tells you how far your final answer is from the accepted value. Together they let you judge a result properly instead of guessing.

  • Years 9 to 12
  • Physics, Chemistry, Biology
  • Updated November 2026

The short answer

Absolute uncertainty

How far either way the true value could be.

Same units as the measurement, written like (1.51 ± 0.04) s.

Percentage uncertainty

The uncertainty as a percentage of the measurement.

Lets you compare how precise different measurements are.

Percentage error

How far the result is from the accepted value.

Only possible when you know the accepted value. It measures accuracy.

Uncertainty from a single reading

The smallest uncertainty any reading can have depends on the scale of the instrument, called its limit of reading.

InstrumentRuleExample
Analogue scale (ruler, thermometer, measuring cylinder)Half the smallest divisionThermometer marked every 1 °C: ± 0.5 °C
Digital display (balance, stopwatch, multimeter)One unit of the last digitBalance reading to 0.01 g: ± 0.01 g
Burette marked every 0.1 mLHalf the smallest division± 0.05 mL per reading

This is the minimum. The real uncertainty is often bigger, for example when you time something by hand or when a reading flickers.

Uncertainty from repeat readings

When you have repeats, the spread of the results usually tells you more than the instrument does. For fewer than 10 readings, use half the range:

uncertainty = ½ × (largest value − smallest value)

Worked example: timing a pendulum

Physics

Four timings of one swing: 1.52 s, 1.48 s, 1.55 s and 1.50 s.

Mean
(1.52 + 1.48 + 1.55 + 1.50) ÷ 4 = 1.5125 s, which rounds to 1.51 s
Uncertainty
Range = 1.55 − 1.48 = 0.07 s. Half the range = 0.035 s, which rounds to 0.04 s
Result
(1.51 ± 0.04) s. Round the uncertainty to one significant figure, then round the value to the same decimal place.

With 10 or more readings, the standard deviation is a better estimate of the uncertainty. Your calculator or a spreadsheet can work it out.

Percentage uncertainty

percentage uncertainty = absolute uncertaintymeasured value × 100%

For the pendulum: 0.04 ÷ 1.51 × 100% = 2.6%. A height of 164 cm measured to ± 1 cm has a percentage uncertainty of only 0.6%, even though 1 cm is a much bigger number than 0.04 s. Percentages let you compare the two fairly.

The same absolute uncertainty matters more for a small measurement. Timing 10 swings instead of one gives a time ten times longer with about the same reaction-time uncertainty, so the percentage uncertainty is about ten times smaller.

Combining uncertainties in a calculation

If you…Then…Example
Add or subtractAdd the absolute uncertaintiesTemperature rise: (47.5 ± 0.5) − (21.0 ± 0.5) = (26.5 ± 1.0) °C
Multiply or divideAdd the percentage uncertaintiesDensity: see below
Raise to a powerMultiply the percentage uncertainty by the powerT = 1.51 s ± 2.6%, so T² = 2.28 s² ± 5.3%, which is ± 0.12 s²

Worked example: density of a metal block

Physics and Chemistry

Mass = (27.0 ± 0.1) g. Volume = (10.0 ± 0.2) cm³. Density = mass ÷ volume.

Value
27.0 ÷ 10.0 = 2.70 g/cm³
Percentage uncertainties
Mass: 0.1 ÷ 27.0 × 100% = 0.4%. Volume: 0.2 ÷ 10.0 × 100% = 2.0%. Total: 0.4% + 2.0% = 2.4%
Result
2.4% of 2.70 = 0.06, so density = (2.70 ± 0.06) g/cm³. The accepted value for aluminium, 2.70 g/cm³, lies inside this range.

Percentage error

percentage error = | experimental value − accepted value |accepted value × 100%

Example: a class finds g = 9.63 m/s². The accepted value is 9.80 m/s². Percentage error = 0.17 ÷ 9.80 × 100% = 1.7%.

There is no fixed percentage error that counts as “accurate enough”. What a good result looks like depends on the experiment and its uncertainty.

Common mistakes

Mixing up uncertainty and error.

Instead: Uncertainty is the doubt in your measurement. Error is the difference from the accepted value.

Giving an uncertainty like ± 0.0347 s.

Instead: Round the uncertainty to one significant figure, then round the value to match.

Adding absolute uncertainties when multiplying.

Instead: For × and ÷ add the percentage uncertainties.

Forgetting units on the uncertainty.

Instead: Write (1.51 ± 0.04) s, with the unit applying to both.

Treating 5% as a pass mark.

Instead: Compare the percentage error with your percentage uncertainty instead.

Exam-style questions

  1. State the uncertainty for (a) a thermometer marked in 1 °C divisions and (b) a digital balance that reads to 0.01 g. 2 marks

    Show a model answer

    (a) ± 0.5 °C, half the smallest division. (b) ± 0.01 g, one unit of the last digit.

  2. A time is measured as (2.0 ± 0.1) s. Calculate the percentage uncertainty. 1 mark

    Show a model answer

    0.1 ÷ 2.0 × 100% = 5%.

  3. The starting temperature of some water is (19.5 ± 0.5) °C and the final temperature is (31.0 ± 0.5) °C. Calculate the temperature change and its uncertainty. 2 marks

    Show a model answer

    31.0 − 19.5 = 11.5 °C. When subtracting, add the absolute uncertainties: 0.5 + 0.5 = 1.0 °C. Temperature change = (11.5 ± 1.0) °C.

  4. A student finds the concentration of a solution to be 0.95 mol/L. Its known concentration is 1.00 mol/L. Calculate the percentage error. 1 mark

    Show a model answer

    |0.95 − 1.00| ÷ 1.00 × 100% = 5.0%.

  5. A student obtains g = (9.62 ± 0.05) m/s². The accepted value is 9.80 m/s². Assess whether the result suggests a systematic error. 3 marks

    Show a model answer

    The uncertainty range is 9.57 to 9.67 m/s², which does not include 9.80 m/s². The difference between the result and the accepted value is larger than the uncertainty can explain, so a systematic error is likely, such as swinging the pendulum through too large an angle.

Questions students and parents ask

What is the difference between uncertainty and error?

Uncertainty is the range within which the true value probably lies, based on your measurements. Error is how far your result actually is from the accepted value. You can work out uncertainty without knowing the true value, but not error.

What is a good percentage error?

There isn’t one fixed number. A good result is one where the accepted value lies within your uncertainty range. A 3% error can be excellent for one experiment and poor for another.

Why is the uncertainty half the smallest division?

On an analogue scale you can usually tell which mark the reading is closest to, so you can’t be out by more than half a division. A digital display rounds in a way you can’t see, so its uncertainty is a whole unit of the last digit.

How many significant figures should an uncertainty have?

Usually one. Then round the measured value to the same decimal place as the uncertainty, for example (1.51 ± 0.04) s.

Keep going