Add Percentage Errors for Products and Quotients, Multiplying by the Power
Every measurement has an uncertainty, and a calculated quantity inherits uncertainty from each input. The rules for combining them are short: add absolute errors for sums and differences, add percentage errors (each multiplied by its power) for products, quotients and powers.
Measurement, Significant Figures and Error Propagation is the central idea in this question pattern. Identify the condition in the stem before applying the method below.
Reading instruments and significant figures
Leading zeros are not significant, while trailing zeros after a decimal point are: 0.007 has 1 significant figure, 2.64 x 10²⁴ has 3, 0.0006032 has 4 and 6.3200 has 5. A stopwatch of least count 1/5 s measuring 25 s has a maximum error of 0.2/25 = 0.8%. For a screw gauge the reading is MSR + (circular reading x LC) minus the zero error, with its sign: 5 mm + 25 x 0.001 cm = 0.525 cm, and a zero error of -0.004 cm means subtracting a negative number, which adds 0.004 cm, giving 0.529 cm.
Mean values and absolute errors
The mean of 1.34, 1.38, 1.32 and 1.36 is 1.35, the deviations are 0.01, 0.03, 0.03 and 0.01 so the mean absolute error is 0.02, and the percentage error is 0.02 / 1.35 = 1.48%. For periods 2.63, 2.56, 2.42, 2.71 and 2.80 s the mean is 2.624 s and the mean of the absolute deviations is about 0.11 s. Many readings reduce random errors, but they cannot remove a systematic error such as defective weights or a zero error.
Propagation rules
OlympIQ exam lens
- ✓Count significant figures: leading zeros no, trailing zeros after a decimal point yes
- ✓Least-count error = LC / reading; screw gauge = MSR + CSR x LC - zero error (with sign)
- ✓Sums and differences: add absolute errors; products and quotients: add percentage errors
- ✓Powers: multiply the percentage error by the exponent
- ✓Many readings reduce random error but not systematic error
Do not subtract percentage errors for a quotient. Errors add in both products and quotients, since the worst case is always the largest possible total deviation, so R = V/I carries the sum of the two percentage errors.
Ready to try the pattern?
100% coverage across the 14 linked questions.
Practise 14 linked questions