Key takeaways
- Significant figures describe precision, not the number of decimal places.
- Leading zeros never count. 0.0045 has two significant figures.
- Trailing zeros count only after a decimal point or in scientific notation.
- Multiply and divide: keep the fewest significant figures of any input.
- Add and subtract: keep the fewest decimal places of any input.
- Round at the end, then attach the unit.
What significant figures record
A measurement is never exact. A scale that reads 12.30 g is claiming a precision of two decimal places, while 12.3 g claims only one, even though the two numbers are close in size. Significant figures are the compact way of stating that claim: they describe which digits the instrument actually supports.
The idea is about relative precision, not absolute size. A distance of 1.2 km and a thickness of 1.2 mm both carry two significant figures, but the first is uncertain by about 50 m and the second by about 0.05 mm. That is why significant figures travel with the measurement and are lost when the number is written without its context.
Counting rules
| Value | Significant figures | Reason |
|---|---|---|
| 0.00450 | 3 | Leading zeros are placeholders; the trailing zero counts. |
| 1200 | 2, 3, or 4 | Ambiguous in plain integer form. |
| 1.200 × 10³ | 4 | Scientific notation states the precision. |
| 500.0 | 4 | The decimal point makes trailing zeros significant. |
| 0.0308 | 3 | 3, 0 (between), and 8 count. |
| 6.02 × 10²³ | 3 | Every digit of the coefficient counts. |
The integer case is the one to watch. Writing 1200 without further information leaves the reader unable to tell whether the zeros were measured or merely hold the decimal place, so scientific notation settles it: 1.2 × 10³ has two figures, 1.200 × 10³ has four.
Exact numbers and conversion factors
Some numbers are exact by definition and never limit precision. Counted objects are exact, so “12 eggs” has no uncertainty attached. Defined conversion factors are exact as well: 1 inch is exactly 2.54 cm, and 1 minute is exactly 60 seconds, because those relations define the units rather than measure them.
Treating a conversion factor as a measured value is a common reason for over-rounding. Converting 8.5 inches to centimetres gives 8.5 × 2.54 = 21.59 cm, which rounds to 22 cm because 8.5 has two significant figures. The factor 2.54 contributes no limit of its own.
Rounding without double-rounding
Round once, at the end. Rounding 2.349 first to 2.35 and then to 2.4 gives a different result from rounding 2.349 directly to 2.3, and the extra step silently discards information. When a value lands exactly halfway, conventions differ: round-half-up sends 2.45 to 2.5, while round-half-even sends it to 2.4 because 4 is even.
Binary floating point complicates the tie. A number typed as 2.675 is often stored slightly below the true value, so software may round it to 2.67 rather than 2.68. Spreadsheet ROUND functions generally round halves away from zero; statistical software may use half-even by default. For published data, state the rule you used.
Multiplication and division
For products and quotients, the answer carries the same number of significant figures as the least precise factor.
4.56 × 1.4 = 6.384 → 6.4 (two significant figures)
The raw product has four digits, but 1.4 has only two, so the digits beyond the first decimal place in 6.384 are not justified. The same rule applies to division: 8.45 g ÷ 2.5 mL = 3.38 g/mL, which becomes 3.4 g/mL. Intermediate results keep at least one extra digit so that rounding does not accumulate.
Addition and subtraction
For sums and differences, the answer keeps the fewest decimal places of any input, because that is where the absolute uncertainty lies.
12.11 + 0.3 = 12.41 → 12.4 (one decimal place)
Here the significant-figure count is irrelevant: 0.3 is known only to a tenth, so the hundredths digit in 12.41 is noise. A large integer can behave the same way. 1,200 + 3.4 rounds to 1,200 when the hundreds digit is the last reliable one, which is another argument for writing such values in scientific notation.
The rule also explains a laboratory surprise. Adding 0.05 g to 250 g gives 250.05 g, but if the balance reads to the nearest gram the sum is reported as 250 g, because the small addition is invisible at that precision. Writing 250.05 g would claim a hundredth of a gram the instrument never measured.
Worked example
A sample has a mass of 24.36 g and a volume of 9.2 mL. Density is mass ÷ volume = 24.36 ÷ 9.2 = 2.6478 g/mL. Mass has four significant figures and volume has two, so the quotient keeps two: 2.6 g/mL.
Now chain two operations. The temperature rises from 12.34 °C to 12.58 °C, a change of 0.24 °C, and the energy supplied is 12.5 J. Heat capacity is 12.5 ÷ 0.24 = 52.08 J/°C. The change has two significant figures (0.24) and the energy has three, so the answer is 52 J/°C. Rounding the temperature change to one decimal place before dividing would have given 0.2 °C and a very different result, which is why the subtraction rule must be applied to the raw values first.
Scientific notation and unambiguous precision
Scientific notation removes the ambiguity in trailing zeros. Writing 4.50 × 10² states three significant figures, while 4.5 × 10² states two and 450 states nothing definite. The coefficient carries the precision and the power of ten carries the scale.
The convention also keeps arithmetic readable. Multiplying 3.0 × 10⁴ by 2.0 × 10³ multiplies the coefficients and adds the exponents, giving 6.0 × 10⁷ with two significant figures, consistent with the inputs. Keeping the coefficient between 1 and 10 makes the digit count obvious at a glance.
Rounding inside scientific notation follows the same rules on the coefficient. 6.28318 becomes 6.28 at three significant figures and 6.3 at two, and the exponent changes only when the coefficient rounds up to 10, in which case 9.98 × 10² becomes 1.0 × 10³.
Common mistakes
- Counting leading zeros. 0.0025 has two significant figures, not five.
- Assuming integer trailing zeros are significant. 1500 is ambiguous; write 1.5 × 10³ or 1.500 × 10³.
- Using the multiplication rule for addition. Sums follow decimal places, not digit counts.
- Rounding twice. Carry extra digits through the calculation and round once at the end.
- Copying every digit from a calculator. A display with ten digits does not create ten digits of precision.
- Applying significant figures to exact counts. Twelve trials are twelve, not one or two significant figures.
Practical caveats
Significant figures are a rough bookkeeping device, not a full uncertainty analysis. Two measurements with the same digit count can have very different error distributions, and a value near a rounding boundary can shift by a full unit in the last place. Where the result matters, report an explicit uncertainty, such as 2.65 ± 0.05 g/mL, which says more than the digit count alone.
Software adds another layer. Floating-point arithmetic cannot represent most decimal fractions exactly, so 0.1 + 0.2 evaluates to 0.30000000000000004 in many languages. That is a representation artifact, not a measurement result. Format the output to the precision you can defend rather than the precision the hardware happens to print.
Frequently asked questions
How many significant figures does 0.00450 have?
Three. The leading zeros are placeholders, the 4 and 5 count, and the trailing zero after the decimal point counts.
Are trailing zeros in a whole number significant?
Not reliably. 1200 could mean two, three, or four significant figures, so write it in scientific notation to remove the doubt.
Do I round before or after the calculation?
After. Keep at least one extra digit through intermediate steps and round only the final answer.
How do significant figures work for percentages?
A percentage inherits the precision of the numbers behind it. A ratio of two two-figure measurements is justified to two figures, so 0.62 becomes 62 %.
Is a conversion factor a significant-figure limit?
Usually not, because defined factors such as 2.54 cm per inch are exact. Measured conversion factors, such as a lab-derived calibration, do carry precision.
Why does my calculator show more digits than I need?
The display shows the internal value, not the precision of the inputs. Report only the digits the measurement supports.
Further reading