Uncertainties, errors and significant figures
In a minute
No measurement is exact. You record each one with its uncertainty, as a range such as , and to the right number of . Then you carry them through calculations and compare the result with the accepted value.
You'll be able to
- Use SI units and prefixes, and convert between them.
- Give measured and calculated values to a sensible number of significant figures or decimal places.
- Record a measurement as a range () and tell from .
- Calculate and interpret , , and .
- Propagate uncertainties through a calculation and judge what they mean for the final result.
Before you start2 quick questions on what this builds on
Revise first: 5.2 Limiting reactants, yield and atom economy
SI units, prefixes and significant figures
Chemists use SI units: kilograms, metres, seconds, kelvin and moles. Volumes are usually in or .
A prefix scales a unit by a power of ten: kilo (k) is , centi (c) is , milli (m) is and micro () is . The full list is in the data booklet, section 3.
For areas and volumes, the prefix is squared or cubed too. , so . Likewise (data booklet, section 4).
Try it first, then reveal one step at a time.
- , because each centimetre is metres and a volume has three lengths.
- .
- Keep 3 significant figures, as in . Sense check: a syringe is far smaller than one cubic metre.
So has 3 significant figures: the leading zeros don't count, the final zero does. has 4, because both zeros come at the end of a number with a decimal point. A whole number such as is ambiguous, so write to show 3.
Try it first, then reveal one step at a time.
- (data booklet, section 7).
- . Keep every digit for now.
- , so .
- The mass has 2 significant figures and the volume has 4. Round to the fewer: 2.
Recording uncertainty, and random vs systematic errors
Every reading has an : how far the true value could be from what you wrote. You record it after the value with and the same unit.
means the true reading is expected to lie between and .
For a scale you read by eye, such as a burette or thermometer, the uncertainty is usually half the smallest division. For a digital display it is one unit in the last digit. If the glassware has a tolerance printed on it, use that.
Try it first, then reveal one step at a time.
- Smallest division: . Half of it is .
- Halfway between the marks is .
- The uncertainty ends in the second decimal place, and so does the value.
Errors come in two kinds. A scatters readings unpredictably, sometimes high and sometimes low: judging a colour change, or a reading that flickers.
A shifts every reading the same way: a balance that isn't zeroed, or heat lost from a calorimeter in every trial.
Random errors lower . Repeating and averaging reduces their effect. Systematic errors lower , and averaging does nothing to them: you must change the method or calibrate.
Try it first, then reveal one step at a time.
- Spread: the readings agree closely. Offset: each is too high by the same amount.
- Name the error type, say why averaging fails, then give the fix.
- The three readings are very close together, so the results are precise. Every reading is 0.15 g too high, so the results are not accurate. This is a systematic error. Repeating and averaging does not reduce a systematic error, because every reading is shifted the same way. The student should zero (tare) the balance before weighing.
Percentage uncertainty and percentage error
An uncertainty of is small for a titre but large for a one. To compare them, divide by the value.
The is the absolute uncertainty divided by the value. Multiply it by 100 for the . Neither has a unit.
Here the titres are and uncertain. So, for the same absolute uncertainty, small measurements carry large percentage uncertainties, which is why you use larger masses, volumes and temperature changes where you can.
Give absolute and percentage uncertainties to 1 significant figure, or 2 if you need them for a later step or to keep an instrument's decimal place, as in a titre of .
Try it first, then reveal one step at a time.
- Fractional uncertainty: .
- Percentage: .
- Round to 1 significant figure. For comparison, on the same balance would be only .
Uncertainty describes your measurements. compares your result with the accepted value, such as a data-booklet value.
Try it first, then reveal one step at a time.
- Accepted value: .
- Difference: .
- , which is to 3 significant figures.
- The value is less negative than the booklet's: some energy never reached the water.
Percentage change and percentage difference
tells you how much a value has grown or shrunk, compared with where it started. A positive answer is an increase and a negative one is a decrease.
compares two values when neither is the reference, such as results from two methods. Divide the difference by their mean.
So choose by the question: an accepted value means percentage error, a before and after means percentage change, and two equal-status values mean percentage difference.
Try it first, then reveal one step at a time.
- Initial value , final value .
- .
- The minus sign shows a decrease. Give it to 2 significant figures, as the data have.
Try it first, then reveal one step at a time.
- Difference: .
- Mean: .
- .
- The difference has only 2 significant figures, so give 2.
Propagating uncertainties
Most results are : calculated from several raw readings. Each reading's uncertainty feeds into the result. Working out the result's uncertainty is called .
Below, means the absolute uncertainty in a quantity . (It is not a change, as in .)
In words: for adding and subtracting, add the absolute uncertainties. For multiplying and dividing, add the percentage uncertainties. Uncertainties always add, even when the values are subtracted.
Treat exact numbers and data-booklet values, such as or the specific heat capacity of water, , as having no uncertainty unless the question gives one.
Try it first, then reveal one step at a time.
- Titre: .
- A subtraction, so add the absolute uncertainties: .
- Keep it as rather than , so the titre keeps the burette's second decimal place.
Try it first, then reveal one step at a time.
- , so and .
- Percentage uncertainties: mass ; volume . is a data-booklet value, so treat it as exact.
- comes from a division, so add them: .
- Absolute: , which is to 1 significant figure.
- Round the value to the same decimal place as the uncertainty: the fifth.
The largest percentage uncertainty usually dominates the result. To improve an experiment, make that measurement larger or use a more precise instrument for it; improving the others barely helps.
You can also judge a result. The direct test is the range: if the accepted value lies within your result its absolute uncertainty, random errors can explain the difference; if it lies outside, there must be systematic errors too. Comparing the percentage error with the total percentage uncertainty gives the same verdict when the gap is clear. Near the edge, check the range itself, because the two percentages are measured against different values (the accepted value and your result).
Try it first, then reveal one step at a time.
- Compare the two percentages. The gap is large (19.7% against 4%), so the verdict is clear: the accepted value lies well outside the result's uncertainty range.
- Say what the uncertainty can account for, and what must cause the rest.
- Give a likely systematic error for burning a fuel under a can, and say why repeats won't fix it.
- The percentage error is much larger than the total percentage uncertainty, so the accepted value lies well outside the result's uncertainty range. Random errors in the measurements can only account for about 4%. So the rest must come from systematic errors in the method. For example, heat is lost to the surroundings, and some of the fuel burns incompletely. Repeating the experiment would not remove these errors; the method must be changed, for example by insulating the can.
At HL you also propagate through powers. Raising a value to the power multiplies its percentage uncertainty by .
Try it first, then reveal one step at a time.
- Percentage uncertainty in : .
- Squaring is a power of 2, so double it: .
- This is the same as adding for .
Common mistakes
- Subtracting uncertainties when the values are subtracted. A titre or a temperature change has two readings, so add their uncertainties: .
- Adding absolute uncertainties in a multiplication or division. Convert each to a percentage first, then add the percentages.
- Writing the value and its uncertainty to different decimal places, such as . Write .
- Giving a calculated answer to more significant figures than the data. Round to the least precise datum.
- Saying that repeating an experiment removes a systematic error. Repeats only reduce random error; a systematic error needs a change of method or calibration.
Exam tips
- "Calculate" a processed value with its uncertainty: show each percentage uncertainty, their sum, then the value its absolute uncertainty with units.
- "Evaluate" a result: check whether the accepted value lies within your result its absolute uncertainty (comparing the percentage error with the total percentage uncertainty works when the gap is clear), then say whether random errors alone explain it.
- In a Paper 1B data table, record each reading to the precision of the instrument, with the same decimal places down a column.
Key takeaways
Practise this
48 past-paper questions on Uncertainty and error in the Chemistry HL bank, each with a mark scheme and a worked explanation.