Counting the digits that actually matter in your measurements

Most people learn significant figures once in chemistry class and never think about them again until they're handed a spreadsheet full of lab data and realize their answer is wrong by an order of magnitude. The basic rule is straightforward: zeros at the start of a number don't count, zeros between nonzero digits do count, and trailing zeros only count if there's a decimal point somewhere in the number. That's about it for the standard cases. The problems start when you actually have to work with them.

What Significant Figures And Significant Digits Actually Mean In Practice

When I say 3.40 g, I'm telling you the measurement was made with an instrument precise to the hundredths place. When I say 3.4 g, the instrument only went to the tenths place. They look similar on paper but they represent fundamentally different levels of certainty. This matters because every calculation you do afterward inherits that uncertainty. You can't create precision that wasn't there to begin with. The standard rules for operations are simple enough: For addition and subtraction, the answer gets rounded to the same decimal place as the least precise measurement. Add 12.11 g plus 0.3 g plus 4.258 g and you get 16.668 g, but since 0.3 only goes to the tenths place, your answer is 16.7 g. Not 16.67, not 16.668. Just 16.7. For multiplication and division, the answer gets rounded to the same number of significant figures as the measurement with the fewest of them. Multiply 2.5 cm by 3.42 cm and you get 8.55 cm², which rounds to 8.6 cm² because 2.5 has only two significant figures. Here's where people mess up: they count significant figures on intermediate results and carry those rounded values forward through the rest of a multi-step problem. Don't do that. Keep all the digits in your calculator through every step and only round at the very end. If you round early, you accumulate rounding error and your final answer can drift significantly from the correct one. I've seen grad students lose half a percent off their results on titration calculations simply because they rounded a molarity value to three figures before using it in a second equation.

A specific problem I ran into: I was working with a spectrophotometer that reported absorbance values to four decimal places, like 0.3456. The concentration calculation involved dividing by a path length of 1.0 cm and multiplying by a molar absorptivity coefficient of 12,500 L/(mol·cm). The tricky part was that the molar absorptivity was only known to three significant figures based on the calibration curve, even though it looked like five digits. If you treat 12,500 as having five sig figs you get one answer. Treat it as three and you get a different one. The workaround was to go back to the calibration data and check the standard error on the slope of the regression line, which confirmed the coefficient was really only precise to about three figures. Write it as 1.25 × 10 in your notes so there's no ambiguity.

Edge Cases That Textbooks Usually Skip

Leading zeros are placeholders, not significant. 0.0042 has two significant figures. The zeros just tell you where the decimal point is. Trailing zeros after a decimal are significant. 450.0 has four. The decimal point makes the final zero meaningful. Trailing zeros before a decimal are ambiguous. 450 could be two or three significant figures. This is why scientific notation exists. 4.50 × 10² clearly has three. 4.5 × 10² clearly has two. If someone writes just 450 in a paper, ask them what they meant. Exact numbers have infinite significant figures. Counted objects, defined constants, conversion factors like 100 cm in 1 m — these don't limit your precision. If you measured 3.25 g of something and converted to milligrams, the answer is 3250 mg, and the 1000 conversion factor doesn't reduce the significant figures. The result still has three. Logarithms are a different beast entirely. The number of significant figures in the original value becomes the number of decimal places in the log result. If pH = 3.45, the [H] concentration is 3.5 × 10 M, not 3.5481 × 10. Two decimal places in the pH correspond to two significant figures in the concentration. This trips people up constantly.

The Real Problem With Significant Figures

The whole system is an approximation of real uncertainty. It was designed for hand calculations in an era before everyone carried a computer. In practice, error propagation theory gives you a more honest picture of what your numbers actually mean. If you have a measurement of 10.2 ± 0.1 mL and another of 5.00 ± 0.05 mL, the relative uncertainty in the first is about 1% and in the second it's about 1%. Significant figures will tell you the result of a division has two figures, but it won't tell you whether your answer is really 2.04 ± 0.04 or 2.04 ± 0.08. That requires actual uncertainty analysis. I use significant figures as a quick sanity check, not as my primary error accounting method. For routine lab work where you're doing a few straightforward calculations, sig figs are fine and fast. If you're publishing or doing anything that needs defensible error bars, propagate the uncertainties properly. Modern spreadsheets and tools like Python's uncertainties package make this trivial. The one situation where sig figs completely break down is with logarithmic or exponential relationships, where the rule of thumb changes depending on which operation you're performing. Base-10 logs, natural logs, exponentials, and antilogs each have their own convention, and mixing them up gives you the wrong number of figures without any obvious warning sign.

Bottom line: learn the rules, apply them consistently, but understand they're a rough shorthand. The alternative isn't more precision — it's better tracking of what you actually know.