Significant figures and rounding

A lab technician reports a length as $12.3\ \text{cm}$ . Another reports the same length as $12.30\ \text{cm}$ . To a careless eye these look like the same number with an extra decorative zero. To a physicist they are two different claims about how well the measurement was made — the second instrument could distinguish hundredths of a centimetre and the first could not. That silent difference, carried entirely in how many digits you choose to write down, is what significant figures are for. Every measured number is a compromise between the digits an instrument can genuinely resolve and the temptation of a calculator to spit out ten more digits than the data deserve. Significant figures are the discipline that keeps a result honest: enough digits to use all the certainty you actually have, and not one digit more, because that next digit would be a fiction invented by arithmetic rather than a fact read off an instrument. Get this wrong and a board exam answer sheet loses marks even when the "physics" — the equation, the substitution, the final number — is completely correct, because the last digit claims a precision nobody measured. This article builds the full toolkit CBSE Class 11 expects: how to count significant figures in any number, including the ambiguous whole numbers that trip up nearly everyone; how to round a result correctly, including the special "round half to even" rule for a trailing 5 that catches students by surprise; why switching units must never change the significant-figure count; and how all of this becomes a fast, mechanical exam skill rather than a source of lost marks. What we will not cover here is what happens to significant figures when numbers are added, multiplied or raised to a power — that is uncertainty propagation, the very next topic in this chapter.

Key takeaways

Significant figures = reliably known digits + one estimated digit. They record how precisely a quantity was actually measured, not how many digits a calculator happens to print.
Four counting rules to memorise: all non-zero digits count; zeros sandwiched between non-zero digits count; leading zeros never count (they just place the decimal point); trailing zeros after a decimal point always count.
Whole numbers without a decimal point are ambiguous — "15000" could have anywhere from 2 to 5 significant figures; scientific notation removes the ambiguity completely.
Rounding follows three cases: dropped digit less than 5 stays put, dropped digit more than 5 rounds up, and dropped digit exactly 5 with nothing after it uses "round half to even" to avoid a systematic upward bias.
Scientific notation, $a\times10^{n}$ with $1\le a<10$ , is the gold standard — the digit count of the coefficient $a$ is the significant-figure count, with zero ambiguity.
A physical quantity's significant figures cannot change just by changing units — this is a fast, reliable self-check on every conversion you do.
Key idea: significant figures are the reliably-known digits of a measurement plus one estimated digit, counted by four fixed rules and rounded by three fixed cases (including round-half-to-even), and scientific notation is what removes every ambiguity that plain decimal writing leaves behind.

The core insight

A measured number's significant figures are the digits you can trust plus exactly one uncertain digit, and both counting them (via four rules, with scientific notation resolving ambiguous trailing zeros) and rounding them (via three cases, with round-half-to-even for an exact trailing 5) exist purely to stop a result from claiming more precision than the measurement actually delivered.

Where it fits in the syllabus

This is the precision-and-honesty half of the CBSE Class 11 Physics unit "Units and Measurements." After learning what SI units are, the syllabus turns to the fact that no measurement is perfect, and significant figures are the notation that carries that imperfection along with the number itself — how many digits, and rounded which way. It is prescribed content, tested directly (count-the-digits and round-this-number questions are exam staples), and it is also a prerequisite skill: the very next topic, arithmetic operations and uncertainty propagation, asks how the significant-figure count of a result is fixed by the significant-figure counts of the inputs — a question that only makes sense once counting and rounding are automatic.

The physical picture

A measured quantity is written as a string of digits, and CBSE's rule is simple to state but easy to misapply: every digit that the instrument actually resolved is significant, plus exactly one more digit that is estimated by eye (or by the least count of the instrument) beyond the smallest marked division. Nothing beyond that estimated digit belongs in the number — carrying it forward is inventing precision that was never measured.

\text{number of significant figures} = (\text{reliably known digits}) + (\text{one estimated digit}).

where: "reliably known digits" are the digits you can read directly off the instrument's scale with confidence, and the "one estimated digit" is the extra digit obtained by judging the fraction of the smallest division — for a metre scale marked in millimetres, you read the millimetre mark confidently and then estimate the next digit (tenths of a millimetre) by eye. A reading of $12.30\ \text{cm}$ therefore states four significant figures: three read with confidence and one estimated, and that trailing zero is not decoration — it is a claim that the hundredths digit was estimated as zero, which is real information a plainer " $12.3\ \text{cm}$ " does not carry. The four counting rules that make this operational are worth stating precisely, because CBSE questions are built to test exactly the edge cases:

Rule 1 — all non-zero digits are significant. In $42.3\ \text{cm}$ , the digits 4, 2 and 3 are all significant (3 significant figures).

Rule 2 — zeros between two non-zero digits are significant. In $5.008$ , both zeros sit between the leading 5 and the trailing 8, so all four digits count (4 significant figures) — a "sandwiched" zero is never dropped.

Rule 3 — leading zeros are never significant. In $0.00308$ , the zeros before the 3 exist only to place the decimal point; they carry no information about precision. Counting from the first non-zero digit, $0.00308$ has 3 significant figures (3, 0, 8).

Rule 4 — trailing zeros after a decimal point are always significant. In $3.500$ , all four digits count, because writing the two extra zeros after the decimal point is a deliberate statement that the hundredths and thousandths digits were measured (or estimated) to be zero. The one case the rules leave unresolved on purpose is a whole number with trailing zeros and no decimal point, such as $15000$ . Is the trailing string of zeros information (the value is known to the last digit) or just placeholders (only the leading digits are trustworthy)? Plain decimal notation genuinely cannot say — and this is exactly the ambiguity scientific notation was invented to remove.

Figure

A ladder of examples showing which digits are significant, with leading zeros greyed out, sandwiched and trailing zeros highlighted, and the same value rewritten in scientific notation with an unambiguous digit count

The fix in Rule 5 is scientific notation: write the number as $a\times10^{n}$ with $1\le a<10$ , and the digit count of the coefficient $a$ is, with no further argument possible, the significant-figure count. $4.700\times10^{3}$ unambiguously has 4 significant figures; the ambiguous "4700" does not tell you whether the last two zeros were measured or merely placed. Rule 6 follows immediately and gives a powerful self-check: since scientific notation is unit-independent in this sense, converting a measurement from one unit to another (metres to centimetres, say) must never change how many significant figures it has — if your converted answer suddenly has more or fewer significant digits than the original, the conversion was done wrong, not the physics.

Derivation ARounding a result to the correct number of significant figures

The goal is to state, as a small, exhaustive set of cases, exactly how to drop the digits beyond the last significant figure you are keeping — this is not a physics derivation in the sense of deriving a formula, but it is the logical procedure CBSE expects to be applied mechanically and correctly on every rounding question, and getting even one case wrong (especially the exact-5 case) is the single most common way marks are lost in this topic. There are exactly three cases, based on the value of the first digit being dropped.

Step 1. If the first digit to be dropped is less than 5, the preceding (retained) digit is left unchanged. Example: rounding $45.657$ to 3 significant figures keeps $45.6$ -something and looks at the digit after the third significant figure; here that digit is $5$ , so this case does not apply on its own — the digit after the 5 must also be checked (see Step 3). A clean less-than-5 case: rounding $8.372$ to 3 significant figures — the digit to drop is $2$ , which is less than 5, so the answer is simply $8.37$ .

Step 2. If the first digit to be dropped is more than 5, the preceding retained digit is raised by 1. Example: rounding $6.3251$ to 3 significant figures keeps $6.32$ and examines the next digit, $5$ , followed by a non-zero digit $1$ — this pushes the case into Step 3's "5 followed by non-zero" rule, which also rounds up, giving $6.33$ . A clean more-than-5 case: rounding $45.657$ to 3 significant figures — after keeping $45.6$ , the next digit is $5$ but it is followed by $7$ (non-zero), so treat it as "more than 5" in effect and round up to $45.7$ .

Step 3. If the first digit to be dropped is exactly 5, look at what follows it. If any non-zero digit follows the 5 (as in $45.657\to45.7$ or $6.3251\to6.33$ above), round the preceding digit up — the trailing digits make the value slightly more than "exactly half", so it rounds up like Step 2. But if the 5 is followed by nothing at all, or only by zeros, the value is exactly halfway between two possible roundings, and rounding it up every single time would introduce a small systematic bias across many measurements. CBSE therefore prescribes round half to even: round the preceding digit to whichever neighbour is even. Rounding $8.375$ to 3 significant figures keeps $8.37$ -something with a bare trailing 5; the digit before it, 7, is odd, so it rounds up to the even neighbour 8, giving $8.38$ . Rounding $8.365$ to 3 significant figures keeps $8.36$ -something with a bare trailing 5; the digit before it, 6, is already even, so it is left unchanged, giving $8.36$ .

Step 4. Apply the same three cases regardless of how many significant figures are being kept or where the decimal point sits — the procedure only ever looks at the single digit being dropped first, and (only in the exact-5 case) at whether anything non-zero trails it. This is why rounding is a purely mechanical, rule-following step and not a matter of judgement. Verified against the worked classic examples: $1.35\to1.4$ (5 with nothing after, preceding digit 3 is odd, rounds up to even 4) and $1.25\to1.2$ (5 with nothing after, preceding digit 2 is already even, stays 2) — both consistent with Step 3.

Worked examples

Board level - counting significant figures in four numbers

Question: Determine the number of significant figures in each of the following: (a) $0.02308$ , (b) $6.020$ , (c) $15000$ , (d) $1.500\times10^{4}$ .

Step 1. For (a) $0.02308$ , apply Rule 3: the zeros before the first non-zero digit (the two zeros right after the decimal point) are leading zeros and are not significant — they exist only to fix the position of the decimal point. Counting from the first non-zero digit onward gives 2, 3, 0, 8, so (a) has 4 significant figures.

Step 2. For (b) $6.020$ , apply Rule 1 to the non-zero digits 6 and 2, Rule 2 to the zero sandwiched between 6 and 2, and Rule 4 to the trailing zero after the decimal point. All four digits count, so (b) has 4 significant figures.

Step 3. For (c) $15000$ , there is no decimal point and no scientific notation, so the trailing zeros are ambiguous by Rule 5. Conventionally, and unless a problem states otherwise, only the non-zero digits are taken as reliably significant, giving 2 significant figures (the 1 and the 5) — but flag this explicitly as an assumption in an exam answer, since it is exactly the ambiguity scientific notation exists to remove.

Step 4. For (d) $1.500\times10^{4}$ , the coefficient $1.500$ has digits 1, 5, 0, 0, and by Rule 4 both trailing zeros after the decimal point in the coefficient are significant. So (d) has 4 significant figures — note that (d) is numerically close to $15000$ from part (c), yet completely unambiguous about its precision, which is the entire point of preferring scientific notation.

The answers: (a) 4, (b) 4, (c) 2 (by convention, genuinely ambiguous), (d) 4.

Competitive level - rounding four results to three significant figures under time pressure

Question: A JEE-style problem gives four computed results and asks you to round each to 3 significant figures: (a) $45.657$ , (b) $6.3251$ , (c) $8.375$ , (d) $8.365$ .

Step 1. For (a) $45.657$ , keep the first three significant figures $4,5,6$ and inspect the next digit: it is $5$ , followed by $7$ (non-zero), so by the derivation's Step 3 (5-followed-by-non-zero) round up: 45.7.

Step 2. For (b) $6.3251$ , keep $6,3,2$ and inspect the next digit: it is $5$ , followed by $1$ (non-zero), so again round up by the same rule: 6.33.

Step 3. For (c) $8.375$ , keep $8,3,7$ and inspect the next digit: it is exactly $5$ with nothing after it, so apply round-half-to-even. The preceding digit is $7$ (odd), so round up to the even neighbour: 8.38.

Step 4. For (d) $8.365$ , keep $8,3,6$ and inspect the next digit: it is exactly $5$ with nothing after it. The preceding digit is $6$ (already even), so it stays unchanged: 8.36.

The answers: (a) 45.7, (b) 6.33, (c) 8.38, (d) 8.36. Notice (c) and (d) differ by only $0.01$ in the original numbers but round to different last digits purely because of which side of "even" the preceding digit falls on — this is the exact trap competitive exams set. Timing note: this whole family of four sub-questions should take under ninety seconds once the three rounding cases are memorised — the only step that costs time is correctly identifying whether the dropped digit is exactly 5 with nothing trailing (triggers round-half-to-even) or 5 followed by something non-zero (always rounds up). Do not pause to "double check" the even/odd call; it is a single glance at the preceding digit's last binary bit of parity, and re-deriving it from scratch wastes seconds you don't have.

Common traps

Assuming trailing zeros are never significant. They are significant whenever they appear after a decimal point (Rule 4) — this is the single most frequently missed rule, costing marks on both counting and rounding questions.
Treating "15000" as having 5 significant figures by default. Without a decimal point or scientific notation, trailing zeros in a whole number are ambiguous; the safe convention counts only the non-zero digits, and the safest exam habit is to convert to scientific notation whenever the number of significant figures matters.
Applying "round 5 up" blindly in every case. The round-half-to-even rule applies only when the dropped digit is exactly 5 with no non-zero digits following it — if anything non-zero trails the 5 (like $6.3251$ ), the ordinary round-up rule applies instead, not round-half-to-even.
Forgetting that significant figures must survive a unit conversion unchanged. Converting $2.0\ \text{m}$ (2 significant figures) to centimetres must give $2.0\times10^{2}\ \text{cm}$ , still 2 significant figures — not $200\ \text{cm}$ , which reads as ambiguous or as 3 significant figures. This is a fast, reliable self-check: if a conversion changes the significant-figure count, redo it in scientific notation.
Counting digits in a calculator display as the answer's significant figures. A calculator will happily print 10 digits for $22/7$ ; none of those extra digits are "measured," and copying all of them into a final answer is exactly the precision-inflation significant figures exist to prevent.

Common trapTrailing zeros are not automatically insignificant — plain decimal notation is what is genuinely ambiguous, not the physics, and the fix is always scientific notation

The most damaging misconception in this topic is treating "zero" as a special, always-meaningless digit. A zero can be highly significant (after a decimal point, or sandwiched between non-zero digits) or completely insignificant (as a leading placeholder, or as an ambiguous trailing zero in a bare whole number) depending entirely on where it sits. There is no shortcut that avoids checking position — memorising "zeros don't count" as a blanket rule will lose marks on roughly half of every counting question CBSE sets. When in doubt, rewrite the number in scientific notation: the ambiguity always disappears there, because every digit in the coefficient is, by construction, significant.

You should be able to…

Count the number of significant figures in any decimal number, including ones with leading zeros, sandwiched zeros, and trailing zeros after a decimal point.
Recognise when a whole number's trailing zeros are ambiguous, and rewrite the number in scientific notation to remove that ambiguity.
Apply the three rounding cases correctly — dropped digit less than 5, more than 5, and exactly 5 (including round half to even when nothing trails the 5).
Round a calculated result to a specified number of significant figures under exam time pressure without hesitating over the even/odd check.
Explain, in one sentence, why the number of significant figures reflects the precision of the measuring instrument rather than an arbitrary choice.
Use a unit conversion as a self-check: confirm the significant-figure count is unchanged before and after converting.

Ask the tutor: If the round-half-to-even rule still feels like a coin flip you can't predict, ask the AI tutor to drill you on five more borderline "exactly 5" cases until the even/odd check becomes automatic.

Why does a trailing zero after a decimal point count as significant, but a trailing zero in a whole number like 500 usually doesn't?

A trailing zero after a decimal point is a deliberate act — nobody writes "3.500" by accident; writing the extra zeros is a specific claim that the hundredths and thousandths digits were checked and found to be zero. A trailing zero in a bare whole number like "500" carries no such guarantee, because whoever wrote it may have just been rounding to the nearest hundred, or may genuinely mean exactly 500 — decimal notation alone cannot tell you which. That is precisely the gap scientific notation closes: $5.00\times10^{2}$ says 3 significant figures on purpose, $5\times10^{2}$ says 1.

Is round half to even actually used in real physics, or is it just an exam rule?

It is a real convention, sometimes called "banker's rounding," used specifically because always rounding a bare 5 upward introduces a small but real statistical bias when you round many numbers in sequence (as happens in long calculations or large datasets) — the average of your rounded values creeps slightly above the true average. Rounding to the nearest even digit splits these borderline cases roughly 50-50 between rounding up and down, cancelling the bias over many repetitions. In CBSE Class 11 it appears specifically for the "exactly 5, nothing trailing" case; for every other case (5 followed by a non-zero digit, or any digit other than 5) the ordinary rule applies.

If I measure something with a ruler marked in millimetres, how many significant figures should my answer have?

You confidently read the millimetre mark, and then you estimate one further digit by eye — the fraction of a millimetre the edge falls at. So a ruler reading like $4.6\ \text{cm}$ (with the millimetre digit written as tenths of a centimetre) already contains one estimated digit and is correctly stated to that many significant figures; writing $4.60\ \text{cm}$ would falsely claim you could resolve hundredths of a centimetre, which a millimetre-marked ruler cannot do. The general rule is: read to the smallest division with confidence, then add exactly one more estimated digit — no more, no less.

My calculator gives me 8.3749999996 for a calculation — do I really have to throw away all those digits?

Yes, and this is exactly the discipline significant figures enforce. A calculator computes with far more digits than any physical measurement can justify, because it has no idea how precisely your original numbers were actually known. If your inputs were only good to 3 significant figures, the honest final answer is also only good to about 3 significant figures (the precise rule for how many is the next topic, uncertainty propagation) — everything past that point in the calculator's display is numerical noise, not physics, and reporting it as if it were real precision is a factual overstatement about how well you measured something.