Why Significant Figures Matter
If you measure a table as "1.5 meters" you do not know whether it is exactly 1.500000 meters or 1.4999 or 1.5001. You know it is between 1.45 and 1.55. The written value "1.5" carries that uncertainty implicitly: two significant figures, meaning the last digit is the estimated one.
When you convert 1.5 meters to feet, the mathematical result is 4.921259842519685 feet. But writing all sixteen digits would falsely suggest that your original measurement was accurate to the nanometer. The honest answer is that the table is about 4.9 feet, with an uncertainty of about 0.16 feet.
How to Count Significant Figures
The rules are simple once you know them:
- Non-zero digits are always significant. In 123, all three digits count.
- Zeros between non-zero digits are significant. In 1002, all four digits count.
- Leading zeros are not significant. In 0.0045, only 4 and 5 count (two significant figures).
- Trailing zeros after a decimal point are significant. In 1.200, all four digits count.
- Trailing zeros before a decimal point are ambiguous. In 1200, it is unclear whether the zeros are significant. Scientific notation removes the ambiguity: 1.2 × 10³ has two significant figures, 1.200 × 10³ has four.
Multiplication and Division
When multiplying or dividing, the result has the same number of significant figures as the input with the fewest significant figures.
Example: 3.14 × 2.5. The first value has three significant figures, the second has two. The result (7.85) is rounded to two significant figures: 7.9.
Addition and Subtraction
When adding or subtracting, the result is rounded to the same decimal place as the least precise input.
Example: 1.20 + 3.4 = 4.60. But since 3.4 has only one decimal place, the result is rounded to 4.6.
Applying This to Unit Conversions
Unit conversion is fundamentally a multiplication or division by a fixed factor. If the factor is exact (for example, 1 inch = 2.54 cm), the conversion does not reduce the significant-figure count — the result has exactly as many significant figures as the input.
Examples:
- Input: 1.5 m (2 significant figures). Result: 4.9 ft (2 significant figures).
- Input: 12.30 m (4 significant figures). Result: 40.35 ft (4 significant figures).
- Input: 0.005 m (1 significant figure). Result: 0.016 ft (1 significant figure).
If the conversion factor is approximate, the number of significant figures in the factor may also limit the result.
Rounding Rules
The standard convention for rounding is "round half to even" (also called banker's rounding):
- If the digit after the last kept place is less than 5, round down.
- If the digit after is greater than 5, round up.
- If the digit after is exactly 5 (with nothing after it), round to the nearest even number.
So 2.5 rounds to 2, 3.5 rounds to 4, 4.5 rounds to 4, 5.5 rounds to 6. This avoids the upward bias that "round half up" produces over many operations.
How M-Convert Handles This
M-Convert computes in IEEE 754 double-precision floating-point (about 15–16 significant decimal digits) and displays results with an automatic significant-figure cap. The cap scales with the magnitude of the number: small values receive more decimal places so their significant figures are preserved, and large values receive fewer so trailing digits do not appear falsely precise.
If you need a specific precision — say, exactly 2 decimal places — you can read the displayed value and round it yourself using the rules above. M-Convert's policy is to never round more aggressively than necessary, so that you retain control over the final precision.
Practical Advice
- Scientific and engineering work: Track significant figures rigorously.
- Everyday life: Two or three significant figures are usually enough. "1.5 m is about 5 feet" is a better statement than "1.5 m is 4.92126 feet".
- Recipes: Two significant figures (say, 240 mL of milk) are typically sufficient.
- Navigation: Distance to two or three significant figures is enough.
- Billing and finance: Use the precision required by the currency; typically two decimal places, but exact values (like exchange rates) may use more.
When in doubt, keep the input precision and match the output to it.